Hugging Face's logo

Spaces:
arminak6
/
demo
No application file

App Files Community
arminak6 commited on
Commit
e5fbe2d
·
verified ·
1 Parent(s): f337443

Upload folder using huggingface_hub

Browse files
This view is limited to 50 files because it contains too many changes.   See raw diff
Files changed (50) hide show
  1. .gitattributes +3 -0
  2. demo/.config/.last_opt_in_prompt.yaml +1 -0
  3. demo/.config/.last_survey_prompt.yaml +1 -0
  4. demo/.config/.last_update_check.json +1 -0
  5. demo/.config/active_config +1 -0
  6. demo/.config/config_sentinel +0 -0
  7. demo/.config/configurations/config_default +6 -0
  8. demo/.config/default_configs.db +0 -0
  9. demo/.config/gce +1 -0
  10. demo/.config/logs/2024.04.15/13.24.57.449262.log +596 -0
  11. demo/.config/logs/2024.04.15/13.25.24.511550.log +5 -0
  12. demo/.config/logs/2024.04.15/13.25.35.658011.log +169 -0
  13. demo/.config/logs/2024.04.15/13.25.45.199675.log +5 -0
  14. demo/.config/logs/2024.04.15/13.25.59.817323.log +8 -0
  15. demo/.config/logs/2024.04.15/13.26.00.519914.log +8 -0
  16. demo/.gitattributes +38 -0
  17. demo/README.md +6 -0
  18. demo/app.py +7 -0
  19. demo/dust3r/.gitignore +132 -0
  20. demo/dust3r/.gitmodules +3 -0
  21. demo/dust3r/LICENSE +7 -0
  22. demo/dust3r/NOTICE +13 -0
  23. demo/dust3r/README.md +299 -0
  24. demo/dust3r/assets/demo.jpg +0 -0
  25. demo/dust3r/assets/dust3r_archi.jpg +0 -0
  26. demo/dust3r/assets/matching.jpg +0 -0
  27. demo/dust3r/assets/pipeline1.jpg +0 -0
  28. demo/dust3r/checkpoints/DUSt3R_ViTLarge_BaseDecoder_512_dpt.pth +3 -0
  29. demo/dust3r/checkpoints/DUSt3R_ViTLarge_BaseDecoder_512_dpt.pth.1 +3 -0
  30. demo/dust3r/croco/LICENSE +52 -0
  31. demo/dust3r/croco/NOTICE +21 -0
  32. demo/dust3r/croco/README.MD +124 -0
  33. demo/dust3r/croco/assets/Chateau1.png +0 -0
  34. demo/dust3r/croco/assets/Chateau2.png +0 -0
  35. demo/dust3r/croco/assets/arch.jpg +0 -0
  36. demo/dust3r/croco/croco-stereo-flow-demo.ipynb +191 -0
  37. demo/dust3r/croco/datasets/__init__.py +0 -0
  38. demo/dust3r/croco/datasets/crops/README.MD +104 -0
  39. demo/dust3r/croco/datasets/crops/extract_crops_from_images.py +159 -0
  40. demo/dust3r/croco/datasets/habitat_sim/README.MD +76 -0
  41. demo/dust3r/croco/datasets/habitat_sim/__init__.py +0 -0
  42. demo/dust3r/croco/datasets/habitat_sim/generate_from_metadata.py +92 -0
  43. demo/dust3r/croco/datasets/habitat_sim/generate_from_metadata_files.py +27 -0
  44. demo/dust3r/croco/datasets/habitat_sim/generate_multiview_images.py +177 -0
  45. demo/dust3r/croco/datasets/habitat_sim/multiview_habitat_sim_generator.py +390 -0
  46. demo/dust3r/croco/datasets/habitat_sim/pack_metadata_files.py +69 -0
  47. demo/dust3r/croco/datasets/habitat_sim/paths.py +129 -0
  48. demo/dust3r/croco/datasets/pairs_dataset.py +109 -0
  49. demo/dust3r/croco/datasets/transforms.py +95 -0
  50. demo/dust3r/croco/demo.py +55 -0
.gitattributes CHANGED
@@ -36,3 +36,6 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
36
  dust3r/checkpoints/DUSt3R_ViTLarge_BaseDecoder_512_dpt.pth.1 filter=lfs diff=lfs merge=lfs -text
37
  sample_data/mnist_test.csv filter=lfs diff=lfs merge=lfs -text
38
  sample_data/mnist_train_small.csv filter=lfs diff=lfs merge=lfs -text
 
 
 
 
36
  dust3r/checkpoints/DUSt3R_ViTLarge_BaseDecoder_512_dpt.pth.1 filter=lfs diff=lfs merge=lfs -text
37
  sample_data/mnist_test.csv filter=lfs diff=lfs merge=lfs -text
38
  sample_data/mnist_train_small.csv filter=lfs diff=lfs merge=lfs -text
39
+ demo/dust3r/checkpoints/DUSt3R_ViTLarge_BaseDecoder_512_dpt.pth.1 filter=lfs diff=lfs merge=lfs -text
40
+ demo/sample_data/mnist_test.csv filter=lfs diff=lfs merge=lfs -text
41
+ demo/sample_data/mnist_train_small.csv filter=lfs diff=lfs merge=lfs -text
demo/.config/.last_opt_in_prompt.yaml ADDED
@@ -0,0 +1 @@
 
 
1
+ {}
demo/.config/.last_survey_prompt.yaml ADDED
@@ -0,0 +1 @@
 
 
1
+ last_prompt_time: 1713187535.105859
demo/.config/.last_update_check.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"last_update_check_time": 1713187544.7223938, "last_update_check_revision": 20240329151455, "notifications": [], "last_nag_times": {}}
demo/.config/active_config ADDED
@@ -0,0 +1 @@
 
 
1
+ default
demo/.config/config_sentinel ADDED
File without changes
demo/.config/configurations/config_default ADDED
@@ -0,0 +1,6 @@
 
 
 
 
 
 
 
1
+ [component_manager]
2
+ disable_update_check = true
3
+
4
+ [compute]
5
+ gce_metadata_read_timeout_sec = 0
6
+
demo/.config/default_configs.db ADDED
Binary file (12.3 kB). View file
 
demo/.config/gce ADDED
@@ -0,0 +1 @@
 
 
1
+ False
demo/.config/logs/2024.04.15/13.24.57.449262.log ADDED
@@ -0,0 +1,596 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ 2024-04-15 13:25:09,477 DEBUG root Loaded Command Group: ['gcloud', 'components']
2
+ 2024-04-15 13:25:09,481 DEBUG root Loaded Command Group: ['gcloud', 'components', 'update']
3
+ 2024-04-15 13:25:09,483 DEBUG root Running [gcloud.components.update] with arguments: [--allow-no-backup: "True", --compile-python: "True", --quiet: "True", COMPONENT-IDS:7: "['core', 'gcloud-deps', 'bq', 'gcloud', 'gcloud-crc32c', 'gsutil', 'anthoscli']"]
4
+ 2024-04-15 13:25:09,485 INFO ___FILE_ONLY___ Beginning update. This process may take several minutes.
5
+
6
+ 2024-04-15 13:25:09,509 DEBUG urllib3.connectionpool Starting new HTTPS connection (1): dl.google.com:443
7
+ 2024-04-15 13:25:09,581 DEBUG urllib3.connectionpool https://dl.google.com:443 "GET /dl/cloudsdk/channels/rapid/components-2.json HTTP/1.1" 200 214439
8
+ 2024-04-15 13:25:09,598 INFO ___FILE_ONLY___
9
+
10
+ 2024-04-15 13:25:09,598 INFO ___FILE_ONLY___
11
+ Your current Google Cloud CLI version is: 471.0.0
12
+
13
+ 2024-04-15 13:25:09,598 INFO ___FILE_ONLY___ Installing components from version: 471.0.0
14
+
15
+ 2024-04-15 13:25:09,598 INFO ___FILE_ONLY___
16
+
17
+ 2024-04-15 13:25:09,599 DEBUG root Chosen display Format:table[box,title="These components will be removed."](details.display_name:label=Name:align=left,version.version_string:label=Version:align=right,data.size.size(zero="",min=1048576):label=Size:align=right)
18
+ 2024-04-15 13:25:09,600 DEBUG root Chosen display Format:table[box,title="These components will be updated."](details.display_name:label=Name:align=left,version.version_string:label=Version:align=right,data.size.size(zero="",min=1048576):label=Size:align=right)
19
+ 2024-04-15 13:25:09,600 DEBUG root Chosen display Format:table[box,title="These components will be installed."](details.display_name:label=Name:align=left,version.version_string:label=Version:align=right,data.size.size(zero="",min=1048576):label=Size:align=right)
20
+ 2024-04-15 13:25:09,730 INFO ___FILE_ONLY___ ┌─────────────────────────────────────────────────────────────────────────────┐
21
+ 2024-04-15 13:25:09,730 INFO ___FILE_ONLY___
22
+
23
+ 2024-04-15 13:25:09,730 INFO ___FILE_ONLY___ │ These components will be installed. │
24
+ 2024-04-15 13:25:09,730 INFO ___FILE_ONLY___
25
+
26
+ 2024-04-15 13:25:09,730 INFO ___FILE_ONLY___ ├─────────────────────────────────────────────────────┬────────────┬──────────┤
27
+ 2024-04-15 13:25:09,730 INFO ___FILE_ONLY___
28
+
29
+ 2024-04-15 13:25:09,731 INFO ___FILE_ONLY___ │ Name │ Version │ Size │
30
+ 2024-04-15 13:25:09,731 INFO ___FILE_ONLY___
31
+
32
+ 2024-04-15 13:25:09,731 INFO ___FILE_ONLY___ ├─────────────────────────────────────────────────────┼────────────┼──────────┤
33
+ 2024-04-15 13:25:09,731 INFO ___FILE_ONLY___
34
+
35
+ 2024-04-15 13:25:09,731 INFO ___FILE_ONLY___ │
36
+ 2024-04-15 13:25:09,731 INFO ___FILE_ONLY___ BigQuery Command Line Tool
37
+ 2024-04-15 13:25:09,731 INFO ___FILE_ONLY___
38
+ 2024-04-15 13:25:09,731 INFO ___FILE_ONLY___ │
39
+ 2024-04-15 13:25:09,731 INFO ___FILE_ONLY___ 2.1.3
40
+ 2024-04-15 13:25:09,731 INFO ___FILE_ONLY___
41
+ 2024-04-15 13:25:09,731 INFO ___FILE_ONLY___ │
42
+ 2024-04-15 13:25:09,731 INFO ___FILE_ONLY___ 1.7 MiB
43
+ 2024-04-15 13:25:09,731 INFO ___FILE_ONLY___
44
+ 2024-04-15 13:25:09,732 INFO ___FILE_ONLY___ │
45
+ 2024-04-15 13:25:09,732 INFO ___FILE_ONLY___
46
+
47
+ 2024-04-15 13:25:09,732 INFO ___FILE_ONLY___ │
48
+ 2024-04-15 13:25:09,732 INFO ___FILE_ONLY___ BigQuery Command Line Tool (Platform Specific)
49
+ 2024-04-15 13:25:09,732 INFO ___FILE_ONLY___
50
+ 2024-04-15 13:25:09,732 INFO ___FILE_ONLY___ │
51
+ 2024-04-15 13:25:09,732 INFO ___FILE_ONLY___ 2.0.101
52
+ 2024-04-15 13:25:09,732 INFO ___FILE_ONLY___
53
+ 2024-04-15 13:25:09,732 INFO ___FILE_ONLY___ │
54
+ 2024-04-15 13:25:09,732 INFO ___FILE_ONLY___ < 1 MiB
55
+ 2024-04-15 13:25:09,732 INFO ___FILE_ONLY___
56
+ 2024-04-15 13:25:09,732 INFO ___FILE_ONLY___ │
57
+ 2024-04-15 13:25:09,732 INFO ___FILE_ONLY___
58
+
59
+ 2024-04-15 13:25:09,732 INFO ___FILE_ONLY___ │
60
+ 2024-04-15 13:25:09,733 INFO ___FILE_ONLY___ Bundled Python 3.11
61
+ 2024-04-15 13:25:09,733 INFO ___FILE_ONLY___
62
+ 2024-04-15 13:25:09,733 INFO ___FILE_ONLY___ │
63
+ 2024-04-15 13:25:09,733 INFO ___FILE_ONLY___ 3.11.8
64
+ 2024-04-15 13:25:09,733 INFO ___FILE_ONLY___
65
+ 2024-04-15 13:25:09,733 INFO ___FILE_ONLY___ │
66
+ 2024-04-15 13:25:09,733 INFO ___FILE_ONLY___ 74.9 MiB
67
+ 2024-04-15 13:25:09,733 INFO ___FILE_ONLY___
68
+ 2024-04-15 13:25:09,733 INFO ___FILE_ONLY___ │
69
+ 2024-04-15 13:25:09,733 INFO ___FILE_ONLY___
70
+
71
+ 2024-04-15 13:25:09,733 INFO ___FILE_ONLY___ │
72
+ 2024-04-15 13:25:09,733 INFO ___FILE_ONLY___ Cloud Storage Command Line Tool
73
+ 2024-04-15 13:25:09,733 INFO ___FILE_ONLY___
74
+ 2024-04-15 13:25:09,734 INFO ___FILE_ONLY___ │
75
+ 2024-04-15 13:25:09,734 INFO ___FILE_ONLY___ 5.27
76
+ 2024-04-15 13:25:09,734 INFO ___FILE_ONLY___
77
+ 2024-04-15 13:25:09,734 INFO ___FILE_ONLY___ │
78
+ 2024-04-15 13:25:09,734 INFO ___FILE_ONLY___ 11.3 MiB
79
+ 2024-04-15 13:25:09,734 INFO ___FILE_ONLY___
80
+ 2024-04-15 13:25:09,734 INFO ___FILE_ONLY___ │
81
+ 2024-04-15 13:25:09,734 INFO ___FILE_ONLY___
82
+
83
+ 2024-04-15 13:25:09,734 INFO ___FILE_ONLY___ │
84
+ 2024-04-15 13:25:09,734 INFO ___FILE_ONLY___ Cloud Storage Command Line Tool (Platform Specific)
85
+ 2024-04-15 13:25:09,734 INFO ___FILE_ONLY___
86
+ 2024-04-15 13:25:09,734 INFO ___FILE_ONLY___ │
87
+ 2024-04-15 13:25:09,734 INFO ___FILE_ONLY___ 5.27
88
+ 2024-04-15 13:25:09,734 INFO ___FILE_ONLY___
89
+ 2024-04-15 13:25:09,735 INFO ___FILE_ONLY___ │
90
+ 2024-04-15 13:25:09,735 INFO ___FILE_ONLY___ < 1 MiB
91
+ 2024-04-15 13:25:09,735 INFO ___FILE_ONLY___
92
+ 2024-04-15 13:25:09,735 INFO ___FILE_ONLY___ │
93
+ 2024-04-15 13:25:09,735 INFO ___FILE_ONLY___
94
+
95
+ 2024-04-15 13:25:09,735 INFO ___FILE_ONLY___ │
96
+ 2024-04-15 13:25:09,735 INFO ___FILE_ONLY___ Google Cloud CLI Core Libraries (Platform Specific)
97
+ 2024-04-15 13:25:09,735 INFO ___FILE_ONLY___
98
+ 2024-04-15 13:25:09,735 INFO ___FILE_ONLY___ │
99
+ 2024-04-15 13:25:09,735 INFO ___FILE_ONLY___ 2024.01.06
100
+ 2024-04-15 13:25:09,735 INFO ___FILE_ONLY___
101
+ 2024-04-15 13:25:09,735 INFO ___FILE_ONLY___ │
102
+ 2024-04-15 13:25:09,735 INFO ___FILE_ONLY___ < 1 MiB
103
+ 2024-04-15 13:25:09,735 INFO ___FILE_ONLY___
104
+ 2024-04-15 13:25:09,736 INFO ___FILE_ONLY___ │
105
+ 2024-04-15 13:25:09,736 INFO ___FILE_ONLY___
106
+
107
+ 2024-04-15 13:25:09,736 INFO ___FILE_ONLY___ │
108
+ 2024-04-15 13:25:09,736 INFO ___FILE_ONLY___ Google Cloud CRC32C Hash Tool
109
+ 2024-04-15 13:25:09,736 INFO ___FILE_ONLY___
110
+ 2024-04-15 13:25:09,736 INFO ___FILE_ONLY___ │
111
+ 2024-04-15 13:25:09,736 INFO ___FILE_ONLY___ 1.0.0
112
+ 2024-04-15 13:25:09,736 INFO ___FILE_ONLY___
113
+ 2024-04-15 13:25:09,736 INFO ___FILE_ONLY___ │
114
+ 2024-04-15 13:25:09,736 INFO ___FILE_ONLY___ 1.2 MiB
115
+ 2024-04-15 13:25:09,736 INFO ___FILE_ONLY___
116
+ 2024-04-15 13:25:09,736 INFO ___FILE_ONLY___ │
117
+ 2024-04-15 13:25:09,736 INFO ___FILE_ONLY___
118
+
119
+ 2024-04-15 13:25:09,736 INFO ___FILE_ONLY___ │
120
+ 2024-04-15 13:25:09,737 INFO ___FILE_ONLY___ anthoscli
121
+ 2024-04-15 13:25:09,737 INFO ___FILE_ONLY___
122
+ 2024-04-15 13:25:09,737 INFO ___FILE_ONLY___ │
123
+ 2024-04-15 13:25:09,737 INFO ___FILE_ONLY___ 0.2.48
124
+ 2024-04-15 13:25:09,737 INFO ___FILE_ONLY___
125
+ 2024-04-15 13:25:09,737 INFO ___FILE_ONLY___ │
126
+ 2024-04-15 13:25:09,737 INFO ___FILE_ONLY___ 68.9 MiB
127
+ 2024-04-15 13:25:09,737 INFO ___FILE_ONLY___
128
+ 2024-04-15 13:25:09,737 INFO ___FILE_ONLY___ │
129
+ 2024-04-15 13:25:09,737 INFO ___FILE_ONLY___
130
+
131
+ 2024-04-15 13:25:09,737 INFO ___FILE_ONLY___ │
132
+ 2024-04-15 13:25:09,737 INFO ___FILE_ONLY___ gcloud cli dependencies
133
+ 2024-04-15 13:25:09,737 INFO ___FILE_ONLY___
134
+ 2024-04-15 13:25:09,738 INFO ___FILE_ONLY___ │
135
+ 2024-04-15 13:25:09,738 INFO ___FILE_ONLY___ 2021.04.16
136
+ 2024-04-15 13:25:09,738 INFO ___FILE_ONLY___
137
+ 2024-04-15 13:25:09,738 INFO ___FILE_ONLY___ │
138
+ 2024-04-15 13:25:09,738 INFO ___FILE_ONLY___ < 1 MiB
139
+ 2024-04-15 13:25:09,738 INFO ___FILE_ONLY___
140
+ 2024-04-15 13:25:09,738 INFO ___FILE_ONLY___ │
141
+ 2024-04-15 13:25:09,738 INFO ___FILE_ONLY___
142
+
143
+ 2024-04-15 13:25:09,738 INFO ___FILE_ONLY___ └─────────────────────────────────────────────────────┴────────────┴──────────┘
144
+ 2024-04-15 13:25:09,738 INFO ___FILE_ONLY___
145
+
146
+ 2024-04-15 13:25:09,738 INFO ___FILE_ONLY___
147
+
148
+ 2024-04-15 13:25:09,743 DEBUG urllib3.connectionpool Starting new HTTPS connection (1): dl.google.com:443
149
+ 2024-04-15 13:25:09,806 DEBUG urllib3.connectionpool https://dl.google.com:443 "GET /dl/cloudsdk/channels/rapid/RELEASE_NOTES HTTP/1.1" 200 1186568
150
+ 2024-04-15 13:25:09,879 INFO ___FILE_ONLY___ For the latest full release notes, please visit:
151
+ https://cloud.google.com/sdk/release_notes
152
+
153
+
154
+ 2024-04-15 13:25:09,882 INFO ___FILE_ONLY___ ╔═════════════════════════════════════════���══════════════════╗
155
+
156
+ 2024-04-15 13:25:09,882 INFO ___FILE_ONLY___ ╠═ Creating update staging area ═╣
157
+
158
+ 2024-04-15 13:25:09,882 INFO ___FILE_ONLY___ ╚
159
+ 2024-04-15 13:25:09,882 INFO ___FILE_ONLY___ ══════
160
+ 2024-04-15 13:25:09,883 INFO ___FILE_ONLY___ ══════
161
+ 2024-04-15 13:25:09,883 INFO ___FILE_ONLY___ ══════
162
+ 2024-04-15 13:25:10,313 INFO ___FILE_ONLY___ ═
163
+ 2024-04-15 13:25:10,364 INFO ___FILE_ONLY___ ═
164
+ 2024-04-15 13:25:10,414 INFO ___FILE_ONLY___ ═
165
+ 2024-04-15 13:25:10,463 INFO ___FILE_ONLY___ ═
166
+ 2024-04-15 13:25:10,504 INFO ___FILE_ONLY___ ═
167
+ 2024-04-15 13:25:10,543 INFO ___FILE_ONLY___ ═
168
+ 2024-04-15 13:25:10,579 INFO ___FILE_ONLY___ ═
169
+ 2024-04-15 13:25:10,619 INFO ___FILE_ONLY___ ═
170
+ 2024-04-15 13:25:10,659 INFO ___FILE_ONLY___ ═
171
+ 2024-04-15 13:25:10,713 INFO ___FILE_ONLY___ ═
172
+ 2024-04-15 13:25:10,804 INFO ___FILE_ONLY___ ═
173
+ 2024-04-15 13:25:10,917 INFO ___FILE_ONLY___ ═
174
+ 2024-04-15 13:25:10,988 INFO ___FILE_ONLY___ ═
175
+ 2024-04-15 13:25:11,056 INFO ___FILE_ONLY___ ═
176
+ 2024-04-15 13:25:11,129 INFO ___FILE_ONLY___ ═
177
+ 2024-04-15 13:25:11,194 INFO ___FILE_ONLY___ ═
178
+ 2024-04-15 13:25:11,252 INFO ___FILE_ONLY___ ═
179
+ 2024-04-15 13:25:11,313 INFO ___FILE_ONLY___ ═
180
+ 2024-04-15 13:25:11,376 INFO ___FILE_ONLY___ ═
181
+ 2024-04-15 13:25:11,444 INFO ___FILE_ONLY___ ═
182
+ 2024-04-15 13:25:11,507 INFO ___FILE_ONLY___ ═
183
+ 2024-04-15 13:25:11,568 INFO ___FILE_ONLY___ ═
184
+ 2024-04-15 13:25:11,642 INFO ___FILE_ONLY___ ═
185
+ 2024-04-15 13:25:11,714 INFO ___FILE_ONLY___ ═
186
+ 2024-04-15 13:25:11,790 INFO ___FILE_ONLY___ ═
187
+ 2024-04-15 13:25:11,865 INFO ___FILE_ONLY___ ═
188
+ 2024-04-15 13:25:11,944 INFO ___FILE_ONLY___ ═
189
+ 2024-04-15 13:25:12,008 INFO ___FILE_ONLY___ ═
190
+ 2024-04-15 13:25:12,074 INFO ___FILE_ONLY___ ═
191
+ 2024-04-15 13:25:12,134 INFO ___FILE_ONLY___ ═
192
+ 2024-04-15 13:25:12,188 INFO ___FILE_ONLY___ ═
193
+ 2024-04-15 13:25:12,236 INFO ___FILE_ONLY___ ═
194
+ 2024-04-15 13:25:12,299 INFO ___FILE_ONLY___ ═
195
+ 2024-04-15 13:25:12,358 INFO ___FILE_ONLY___ ═
196
+ 2024-04-15 13:25:12,415 INFO ___FILE_ONLY___ ═
197
+ 2024-04-15 13:25:12,465 INFO ___FILE_ONLY___ ═
198
+ 2024-04-15 13:25:12,551 INFO ___FILE_ONLY___ ═
199
+ 2024-04-15 13:25:12,613 INFO ___FILE_ONLY___ ═
200
+ 2024-04-15 13:25:12,695 INFO ___FILE_ONLY___ ═
201
+ 2024-04-15 13:25:12,859 INFO ___FILE_ONLY___ ═
202
+ 2024-04-15 13:25:12,913 INFO ___FILE_ONLY___ ═
203
+ 2024-04-15 13:25:12,971 INFO ___FILE_ONLY___ ═
204
+ 2024-04-15 13:25:12,971 INFO ___FILE_ONLY___ ╝
205
+
206
+ 2024-04-15 13:25:13,063 INFO ___FILE_ONLY___ ╔════════════════════════════════════════════════════════════╗
207
+
208
+ 2024-04-15 13:25:13,063 INFO ___FILE_ONLY___ ╠═ Installing: BigQuery Command Line Tool ═╣
209
+
210
+ 2024-04-15 13:25:13,063 INFO ___FILE_ONLY___ ╚
211
+ 2024-04-15 13:25:13,067 DEBUG urllib3.connectionpool Starting new HTTPS connection (1): dl.google.com:443
212
+ 2024-04-15 13:25:13,126 DEBUG urllib3.connectionpool https://dl.google.com:443 "GET /dl/cloudsdk/channels/rapid/components/google-cloud-sdk-bq-20240329151455.tar.gz HTTP/1.1" 200 1743971
213
+ 2024-04-15 13:25:13,138 INFO ___FILE_ONLY___ ═
214
+ 2024-04-15 13:25:13,138 INFO ___FILE_ONLY___ ═
215
+ 2024-04-15 13:25:13,138 INFO ___FILE_ONLY___ ═
216
+ 2024-04-15 13:25:13,139 INFO ___FILE_ONLY___ ═
217
+ 2024-04-15 13:25:13,139 INFO ___FILE_ONLY___ ═
218
+ 2024-04-15 13:25:13,139 INFO ___FILE_ONLY___ ═
219
+ 2024-04-15 13:25:13,139 INFO ___FILE_ONLY___ ═
220
+ 2024-04-15 13:25:13,139 INFO ___FILE_ONLY___ ═
221
+ 2024-04-15 13:25:13,139 INFO ___FILE_ONLY___ ═
222
+ 2024-04-15 13:25:13,140 INFO ___FILE_ONLY___ ═
223
+ 2024-04-15 13:25:13,140 INFO ___FILE_ONLY___ ═
224
+ 2024-04-15 13:25:13,140 INFO ___FILE_ONLY___ ═
225
+ 2024-04-15 13:25:13,140 INFO ___FILE_ONLY___ ═
226
+ 2024-04-15 13:25:13,140 INFO ___FILE_ONLY___ ═
227
+ 2024-04-15 13:25:13,140 INFO ___FILE_ONLY___ ═
228
+ 2024-04-15 13:25:13,141 INFO ___FILE_ONLY___ ═
229
+ 2024-04-15 13:25:13,141 INFO ___FILE_ONLY___ ═
230
+ 2024-04-15 13:25:13,141 INFO ___FILE_ONLY___ ═
231
+ 2024-04-15 13:25:13,141 INFO ___FILE_ONLY___ ═
232
+ 2024-04-15 13:25:13,141 INFO ___FILE_ONLY___ ═
233
+ 2024-04-15 13:25:13,141 INFO ___FILE_ONLY___ ═
234
+ 2024-04-15 13:25:13,142 INFO ___FILE_ONLY___ ═
235
+ 2024-04-15 13:25:13,142 INFO ___FILE_ONLY___ ═
236
+ 2024-04-15 13:25:13,142 INFO ___FILE_ONLY___ ═
237
+ 2024-04-15 13:25:13,142 INFO ___FILE_ONLY___ ═
238
+ 2024-04-15 13:25:13,142 INFO ___FILE_ONLY___ ═
239
+ 2024-04-15 13:25:13,142 INFO ___FILE_ONLY___ ═
240
+ 2024-04-15 13:25:13,142 INFO ___FILE_ONLY___ ═
241
+ 2024-04-15 13:25:13,143 INFO ___FILE_ONLY___ ═
242
+ 2024-04-15 13:25:13,143 INFO ___FILE_ONLY___ ═
243
+ 2024-04-15 13:25:13,273 INFO ___FILE_ONLY___ ═
244
+ 2024-04-15 13:25:13,279 INFO ___FILE_ONLY___ ═
245
+ 2024-04-15 13:25:13,284 INFO ___FILE_ONLY___ ═
246
+ 2024-04-15 13:25:13,288 INFO ___FILE_ONLY___ ═
247
+ 2024-04-15 13:25:13,292 INFO ___FILE_ONLY___ ═
248
+ 2024-04-15 13:25:13,297 INFO ___FILE_ONLY___ ═
249
+ 2024-04-15 13:25:13,302 INFO ___FILE_ONLY___ ═
250
+ 2024-04-15 13:25:13,305 INFO ___FILE_ONLY___ ═
251
+ 2024-04-15 13:25:13,311 INFO ___FILE_ONLY___ ═
252
+ 2024-04-15 13:25:13,315 INFO ___FILE_ONLY___ ═
253
+ 2024-04-15 13:25:13,319 INFO ___FILE_ONLY___ ═
254
+ 2024-04-15 13:25:13,323 INFO ___FILE_ONLY___ ═
255
+ 2024-04-15 13:25:13,327 INFO ___FILE_ONLY___ ═
256
+ 2024-04-15 13:25:13,334 INFO ___FILE_ONLY___ ═
257
+ 2024-04-15 13:25:13,338 INFO ___FILE_ONLY___ ═
258
+ 2024-04-15 13:25:13,342 INFO ___FILE_ONLY___ ═
259
+ 2024-04-15 13:25:13,348 INFO ___FILE_ONLY___ ═
260
+ 2024-04-15 13:25:13,353 INFO ___FILE_ONLY___ ═
261
+ 2024-04-15 13:25:13,360 INFO ___FILE_ONLY___ ═
262
+ 2024-04-15 13:25:13,364 INFO ___FILE_ONLY___ ═
263
+ 2024-04-15 13:25:13,370 INFO ___FILE_ONLY___ ═
264
+ 2024-04-15 13:25:13,376 INFO ___FILE_ONLY___ ═
265
+ 2024-04-15 13:25:13,380 INFO ___FILE_ONLY___ ═
266
+ 2024-04-15 13:25:13,384 INFO ___FILE_ONLY___ ═
267
+ 2024-04-15 13:25:13,389 INFO ___FILE_ONLY___ ═
268
+ 2024-04-15 13:25:13,393 INFO ___FILE_ONLY___ ═
269
+ 2024-04-15 13:25:13,398 INFO ___FILE_ONLY___ ═
270
+ 2024-04-15 13:25:13,402 INFO ___FILE_ONLY___ ═
271
+ 2024-04-15 13:25:13,406 INFO ___FILE_ONLY___ ═
272
+ 2024-04-15 13:25:13,411 INFO ___FILE_ONLY___ ═
273
+ 2024-04-15 13:25:13,411 INFO ___FILE_ONLY___ ╝
274
+
275
+ 2024-04-15 13:25:13,429 INFO ___FILE_ONLY___ ╔════════════════════════════════════════════════════════════╗
276
+
277
+ 2024-04-15 13:25:13,429 INFO ___FILE_ONLY___ ╠═ Installing: BigQuery Command Line Tool (Platform Spec... ═╣
278
+
279
+ 2024-04-15 13:25:13,429 INFO ___FILE_ONLY___ ╚
280
+ 2024-04-15 13:25:13,433 DEBUG urllib3.connectionpool Starting new HTTPS connection (1): dl.google.com:443
281
+ 2024-04-15 13:25:13,492 DEBUG urllib3.connectionpool https://dl.google.com:443 "GET /dl/cloudsdk/channels/rapid/components/google-cloud-sdk-bq-nix-20240106004423.tar.gz HTTP/1.1" 200 2026
282
+ 2024-04-15 13:25:13,493 INFO ___FILE_ONLY___ ══════════════════════════════
283
+ 2024-04-15 13:25:13,494 INFO ___FILE_ONLY___ ══════════════════════════════
284
+ 2024-04-15 13:25:13,494 INFO ___FILE_ONLY___ ╝
285
+
286
+ 2024-04-15 13:25:13,503 INFO ___FILE_ONLY___ ╔════════════════════════════════════════════════════════════╗
287
+
288
+ 2024-04-15 13:25:13,503 INFO ___FILE_ONLY___ ╠═ Installing: Bundled Python 3.11 ═╣
289
+
290
+ 2024-04-15 13:25:13,503 INFO ___FILE_ONLY___ ╚
291
+ 2024-04-15 13:25:13,507 DEBUG urllib3.connectionpool Starting new HTTPS connection (1): dl.google.com:443
292
+ 2024-04-15 13:25:13,567 DEBUG urllib3.connectionpool https://dl.google.com:443 "GET /dl/cloudsdk/channels/rapid/components/google-cloud-sdk-bundled-python3-unix-linux-x86_64-20240229170130.tar.gz HTTP/1.1" 200 78486918
293
+ 2024-04-15 13:25:13,861 INFO ___FILE_ONLY___ ═
294
+ 2024-04-15 13:25:13,864 INFO ___FILE_ONLY___ ═
295
+ 2024-04-15 13:25:13,868 INFO ___FILE_ONLY___ ═
296
+ 2024-04-15 13:25:13,871 INFO ___FILE_ONLY___ ═
297
+ 2024-04-15 13:25:13,874 INFO ___FILE_ONLY___ ═
298
+ 2024-04-15 13:25:13,878 INFO ___FILE_ONLY___ ═
299
+ 2024-04-15 13:25:13,881 INFO ___FILE_ONLY___ ═
300
+ 2024-04-15 13:25:13,884 INFO ___FILE_ONLY___ ═
301
+ 2024-04-15 13:25:13,888 INFO ___FILE_ONLY___ ═
302
+ 2024-04-15 13:25:13,891 INFO ___FILE_ONLY___ ═
303
+ 2024-04-15 13:25:13,894 INFO ___FILE_ONLY___ ═
304
+ 2024-04-15 13:25:13,898 INFO ___FILE_ONLY___ ═
305
+ 2024-04-15 13:25:13,901 INFO ___FILE_ONLY___ ═
306
+ 2024-04-15 13:25:13,904 INFO ___FILE_ONLY___ ═
307
+ 2024-04-15 13:25:13,908 INFO ___FILE_ONLY___ ═
308
+ 2024-04-15 13:25:13,911 INFO ___FILE_ONLY___ ═
309
+ 2024-04-15 13:25:13,914 INFO ___FILE_ONLY___ ═
310
+ 2024-04-15 13:25:13,917 INFO ___FILE_ONLY___ ═
311
+ 2024-04-15 13:25:13,920 INFO ___FILE_ONLY___ ═
312
+ 2024-04-15 13:25:13,923 INFO ___FILE_ONLY___ ═
313
+ 2024-04-15 13:25:13,927 INFO ___FILE_ONLY___ ═
314
+ 2024-04-15 13:25:13,930 INFO ___FILE_ONLY___ ═
315
+ 2024-04-15 13:25:13,933 INFO ___FILE_ONLY___ ═
316
+ 2024-04-15 13:25:13,937 INFO ___FILE_ONLY___ ═
317
+ 2024-04-15 13:25:13,940 INFO ___FILE_ONLY___ ═
318
+ 2024-04-15 13:25:13,944 INFO ___FILE_ONLY___ ═
319
+ 2024-04-15 13:25:13,947 INFO ___FILE_ONLY___ ═
320
+ 2024-04-15 13:25:13,951 INFO ___FILE_ONLY___ ═
321
+ 2024-04-15 13:25:13,954 INFO ___FILE_ONLY___ ═
322
+ 2024-04-15 13:25:13,958 INFO ___FILE_ONLY___ ═
323
+ 2024-04-15 13:25:16,206 INFO ___FILE_ONLY___ ═
324
+ 2024-04-15 13:25:16,235 INFO ___FILE_ONLY___ ═
325
+ 2024-04-15 13:25:16,267 INFO ___FILE_ONLY___ ═
326
+ 2024-04-15 13:25:16,310 INFO ___FILE_ONLY___ ═
327
+ 2024-04-15 13:25:16,346 INFO ___FILE_ONLY___ ═
328
+ 2024-04-15 13:25:16,375 INFO ___FILE_ONLY___ ═
329
+ 2024-04-15 13:25:16,403 INFO ___FILE_ONLY___ ═
330
+ 2024-04-15 13:25:16,433 INFO ___FILE_ONLY___ ═
331
+ 2024-04-15 13:25:16,464 INFO ___FILE_ONLY___ ═
332
+ 2024-04-15 13:25:16,494 INFO ___FILE_ONLY___ ═
333
+ 2024-04-15 13:25:16,524 INFO ___FILE_ONLY___ ═
334
+ 2024-04-15 13:25:16,553 INFO ___FILE_ONLY___ ═
335
+ 2024-04-15 13:25:16,586 INFO ___FILE_ONLY___ ═
336
+ 2024-04-15 13:25:16,616 INFO ___FILE_ONLY___ ═
337
+ 2024-04-15 13:25:16,649 INFO ___FILE_ONLY___ ═
338
+ 2024-04-15 13:25:16,681 INFO ___FILE_ONLY___ ═
339
+ 2024-04-15 13:25:16,712 INFO ___FILE_ONLY___ ═
340
+ 2024-04-15 13:25:17,142 INFO ___FILE_ONLY___ ═
341
+ 2024-04-15 13:25:17,183 INFO ___FILE_ONLY___ ═
342
+ 2024-04-15 13:25:17,240 INFO ___FILE_ONLY___ ═
343
+ 2024-04-15 13:25:17,289 INFO ___FILE_ONLY___ ═
344
+ 2024-04-15 13:25:17,458 INFO ___FILE_ONLY___ ═
345
+ 2024-04-15 13:25:17,608 INFO ___FILE_ONLY___ ═
346
+ 2024-04-15 13:25:17,653 INFO ___FILE_ONLY___ ═
347
+ 2024-04-15 13:25:17,699 INFO ___FILE_ONLY___ ═
348
+ 2024-04-15 13:25:17,775 INFO ___FILE_ONLY___ ═
349
+ 2024-04-15 13:25:17,815 INFO ___FILE_ONLY___ ═
350
+ 2024-04-15 13:25:17,866 INFO ___FILE_ONLY___ ═
351
+ 2024-04-15 13:25:19,094 INFO ___FILE_ONLY___ ═
352
+ 2024-04-15 13:25:19,129 INFO ___FILE_ONLY___ ═
353
+ 2024-04-15 13:25:19,129 INFO ___FILE_ONLY___ ╝
354
+
355
+ 2024-04-15 13:25:19,248 INFO ___FILE_ONLY___ ╔════════════════════════════════════════════════════════════╗
356
+
357
+ 2024-04-15 13:25:19,249 INFO ___FILE_ONLY___ ╠═ Installing: Bundled Python 3.11 ═╣
358
+
359
+ 2024-04-15 13:25:19,249 INFO ___FILE_ONLY___ ╚
360
+ 2024-04-15 13:25:19,254 INFO ___FILE_ONLY___ ════════════════════════════════════════════════════════════
361
+ 2024-04-15 13:25:19,255 INFO ___FILE_ONLY___ ╝
362
+
363
+ 2024-04-15 13:25:19,257 INFO ___FILE_ONLY___ ╔════════════════════════════════════════════════════════════╗
364
+
365
+ 2024-04-15 13:25:19,257 INFO ___FILE_ONLY___ ╠═ Installing: Cloud Storage Command Line Tool ═╣
366
+
367
+ 2024-04-15 13:25:19,257 INFO ___FILE_ONLY___ ╚
368
+ 2024-04-15 13:25:19,261 DEBUG urllib3.connectionpool Starting new HTTPS connection (1): dl.google.com:443
369
+ 2024-04-15 13:25:19,326 DEBUG urllib3.connectionpool https://dl.google.com:443 "GET /dl/cloudsdk/channels/rapid/components/google-cloud-sdk-gsutil-20231025210228.tar.gz HTTP/1.1" 200 11833901
370
+ 2024-04-15 13:25:19,376 INFO ___FILE_ONLY___ ═
371
+ 2024-04-15 13:25:19,377 INFO ___FILE_ONLY___ ═
372
+ 2024-04-15 13:25:19,378 INFO ___FILE_ONLY___ ═
373
+ 2024-04-15 13:25:19,378 INFO ___FILE_ONLY___ ═
374
+ 2024-04-15 13:25:19,379 INFO ___FILE_ONLY___ ═
375
+ 2024-04-15 13:25:19,380 INFO ___FILE_ONLY___ ═
376
+ 2024-04-15 13:25:19,380 INFO ___FILE_ONLY___ ═
377
+ 2024-04-15 13:25:19,381 INFO ___FILE_ONLY___ ═
378
+ 2024-04-15 13:25:19,381 INFO ___FILE_ONLY___ ═
379
+ 2024-04-15 13:25:19,382 INFO ___FILE_ONLY___ ═
380
+ 2024-04-15 13:25:19,383 INFO ___FILE_ONLY___ ═
381
+ 2024-04-15 13:25:19,383 INFO ___FILE_ONLY___ ═
382
+ 2024-04-15 13:25:19,384 INFO ___FILE_ONLY___ ═
383
+ 2024-04-15 13:25:19,385 INFO ___FILE_ONLY___ ═
384
+ 2024-04-15 13:25:19,385 INFO ___FILE_ONLY___ ═
385
+ 2024-04-15 13:25:19,386 INFO ___FILE_ONLY___ ═
386
+ 2024-04-15 13:25:19,386 INFO ___FILE_ONLY___ ═
387
+ 2024-04-15 13:25:19,387 INFO ___FILE_ONLY___ ═
388
+ 2024-04-15 13:25:19,388 INFO ___FILE_ONLY___ ═
389
+ 2024-04-15 13:25:19,388 INFO ___FILE_ONLY___ ═
390
+ 2024-04-15 13:25:19,389 INFO ___FILE_ONLY___ ═
391
+ 2024-04-15 13:25:19,389 INFO ___FILE_ONLY___ ═
392
+ 2024-04-15 13:25:19,390 INFO ___FILE_ONLY___ ═
393
+ 2024-04-15 13:25:19,391 INFO ___FILE_ONLY___ ═
394
+ 2024-04-15 13:25:19,391 INFO ___FILE_ONLY___ ═
395
+ 2024-04-15 13:25:19,392 INFO ___FILE_ONLY___ ═
396
+ 2024-04-15 13:25:19,393 INFO ___FILE_ONLY___ ═
397
+ 2024-04-15 13:25:19,393 INFO ___FILE_ONLY___ ═
398
+ 2024-04-15 13:25:19,394 INFO ___FILE_ONLY___ ═
399
+ 2024-04-15 13:25:19,394 INFO ___FILE_ONLY___ ═
400
+ 2024-04-15 13:25:20,132 INFO ___FILE_ONLY___ ═
401
+ 2024-04-15 13:25:20,170 INFO ___FILE_ONLY___ ═
402
+ 2024-04-15 13:25:20,201 INFO ___FILE_ONLY___ ═
403
+ 2024-04-15 13:25:20,232 INFO ___FILE_ONLY___ ═
404
+ 2024-04-15 13:25:20,260 INFO ___FILE_ONLY___ ═
405
+ 2024-04-15 13:25:20,290 INFO ___FILE_ONLY___ ═
406
+ 2024-04-15 13:25:20,310 INFO ___FILE_ONLY___ ═
407
+ 2024-04-15 13:25:20,329 INFO ___FILE_ONLY___ ═
408
+ 2024-04-15 13:25:20,352 INFO ___FILE_ONLY___ ═
409
+ 2024-04-15 13:25:20,373 INFO ___FILE_ONLY___ ═
410
+ 2024-04-15 13:25:20,396 INFO ___FILE_ONLY___ ═
411
+ 2024-04-15 13:25:20,416 INFO ___FILE_ONLY___ ═
412
+ 2024-04-15 13:25:20,448 INFO ___FILE_ONLY___ ═
413
+ 2024-04-15 13:25:20,471 INFO ___FILE_ONLY___ ═
414
+ 2024-04-15 13:25:20,503 INFO ___FILE_ONLY___ ═
415
+ 2024-04-15 13:25:20,533 INFO ___FILE_ONLY___ ═
416
+ 2024-04-15 13:25:20,573 INFO ___FILE_ONLY___ ═
417
+ 2024-04-15 13:25:20,607 INFO ___FILE_ONLY___ ═
418
+ 2024-04-15 13:25:20,628 INFO ___FILE_ONLY___ ═
419
+ 2024-04-15 13:25:20,652 INFO ___FILE_ONLY___ ═
420
+ 2024-04-15 13:25:20,683 INFO ___FILE_ONLY___ ═
421
+ 2024-04-15 13:25:20,705 INFO ___FILE_ONLY___ ═
422
+ 2024-04-15 13:25:20,728 INFO ___FILE_ONLY___ ═
423
+ 2024-04-15 13:25:20,757 INFO ___FILE_ONLY___ ═
424
+ 2024-04-15 13:25:20,778 INFO ___FILE_ONLY___ ═
425
+ 2024-04-15 13:25:20,828 INFO ___FILE_ONLY___ ═
426
+ 2024-04-15 13:25:20,856 INFO ___FILE_ONLY___ ═
427
+ 2024-04-15 13:25:20,882 INFO ___FILE_ONLY___ ═
428
+ 2024-04-15 13:25:20,911 INFO ___FILE_ONLY___ ═
429
+ 2024-04-15 13:25:20,932 INFO ___FILE_ONLY___ ═
430
+ 2024-04-15 13:25:20,932 INFO ___FILE_ONLY___ ╝
431
+
432
+ 2024-04-15 13:25:21,005 INFO ___FILE_ONLY___ ╔════════════════════════════════════════════════════════════╗
433
+
434
+ 2024-04-15 13:25:21,005 INFO ___FILE_ONLY___ ╠═ Installing: Cloud Storage Command Line Tool (Platform... ═╣
435
+
436
+ 2024-04-15 13:25:21,005 INFO ___FILE_ONLY___ ╚
437
+ 2024-04-15 13:25:21,009 DEBUG urllib3.connectionpool Starting new HTTPS connection (1): dl.google.com:443
438
+ 2024-04-15 13:25:21,069 DEBUG urllib3.connectionpool https://dl.google.com:443 "GET /dl/cloudsdk/channels/rapid/components/google-cloud-sdk-gsutil-nix-20240106004423.tar.gz HTTP/1.1" 200 2042
439
+ 2024-04-15 13:25:21,070 INFO ___FILE_ONLY___ ══════════════════════════════
440
+ 2024-04-15 13:25:21,071 INFO ___FILE_ONLY___ ══════════════════════════════
441
+ 2024-04-15 13:25:21,071 INFO ___FILE_ONLY___ ╝
442
+
443
+ 2024-04-15 13:25:21,081 INFO ___FILE_ONLY___ ╔════════════════════════════════════════════════════════════╗
444
+
445
+ 2024-04-15 13:25:21,081 INFO ___FILE_ONLY___ ╠═ Installing: Default set of gcloud commands ═╣
446
+
447
+ 2024-04-15 13:25:21,081 INFO ___FILE_ONLY___ ╚
448
+ 2024-04-15 13:25:21,087 INFO ___FILE_ONLY___ ════════════════════════════════════════════════════════════
449
+ 2024-04-15 13:25:21,087 INFO ___FILE_ONLY___ ╝
450
+
451
+ 2024-04-15 13:25:21,089 INFO ___FILE_ONLY___ ╔════════════════════════════════════════════════════════════╗
452
+
453
+ 2024-04-15 13:25:21,089 INFO ___FILE_ONLY___ ╠═ Installing: Google Cloud CLI Core Libraries (Platform... ═╣
454
+
455
+ 2024-04-15 13:25:21,089 INFO ___FILE_ONLY___ ╚
456
+ 2024-04-15 13:25:21,093 DEBUG urllib3.connectionpool Starting new HTTPS connection (1): dl.google.com:443
457
+ 2024-04-15 13:25:21,157 DEBUG urllib3.connectionpool https://dl.google.com:443 "GET /dl/cloudsdk/channels/rapid/components/google-cloud-sdk-core-nix-20240106004423.tar.gz HTTP/1.1" 200 2410
458
+ 2024-04-15 13:25:21,157 INFO ___FILE_ONLY___ ══════════════════════════════
459
+ 2024-04-15 13:25:21,159 INFO ___FILE_ONLY___ ═══════════════
460
+ 2024-04-15 13:25:21,159 INFO ___FILE_ONLY___ ═══════════════
461
+ 2024-04-15 13:25:21,159 INFO ___FILE_ONLY___ ╝
462
+
463
+ 2024-04-15 13:25:21,168 INFO ___FILE_ONLY___ ╔════════════════════════════════════════════════════════════╗
464
+
465
+ 2024-04-15 13:25:21,168 INFO ___FILE_ONLY___ ╠═ Installing: Google Cloud CRC32C Hash Tool ═╣
466
+
467
+ 2024-04-15 13:25:21,168 INFO ___FILE_ONLY___ ╚
468
+ 2024-04-15 13:25:21,172 DEBUG urllib3.connectionpool Starting new HTTPS connection (1): dl.google.com:443
469
+ 2024-04-15 13:25:21,229 DEBUG urllib3.connectionpool https://dl.google.com:443 "GET /dl/cloudsdk/channels/rapid/components/google-cloud-sdk-gcloud-crc32c-linux-x86_64-20231215195722.tar.gz HTTP/1.1" 200 1287877
470
+ 2024-04-15 13:25:21,240 INFO ___FILE_ONLY___ ═
471
+ 2024-04-15 13:25:21,240 INFO ___FILE_ONLY___ ═
472
+ 2024-04-15 13:25:21,240 INFO ___FILE_ONLY___ ═
473
+ 2024-04-15 13:25:21,241 INFO ___FILE_ONLY___ ═
474
+ 2024-04-15 13:25:21,241 INFO ___FILE_ONLY___ ═
475
+ 2024-04-15 13:25:21,241 INFO ___FILE_ONLY___ ═
476
+ 2024-04-15 13:25:21,241 INFO ___FILE_ONLY___ ═
477
+ 2024-04-15 13:25:21,241 INFO ___FILE_ONLY___ ═
478
+ 2024-04-15 13:25:21,241 INFO ___FILE_ONLY___ ═
479
+ 2024-04-15 13:25:21,241 INFO ___FILE_ONLY___ ═
480
+ 2024-04-15 13:25:21,242 INFO ___FILE_ONLY___ ═
481
+ 2024-04-15 13:25:21,242 INFO ___FILE_ONLY___ ═
482
+ 2024-04-15 13:25:21,242 INFO ___FILE_ONLY___ ═
483
+ 2024-04-15 13:25:21,242 INFO ___FILE_ONLY___ ═
484
+ 2024-04-15 13:25:21,242 INFO ___FILE_ONLY___ ═
485
+ 2024-04-15 13:25:21,242 INFO ___FILE_ONLY___ ═
486
+ 2024-04-15 13:25:21,242 INFO ___FILE_ONLY___ ═
487
+ 2024-04-15 13:25:21,242 INFO ___FILE_ONLY___ ═
488
+ 2024-04-15 13:25:21,243 INFO ___FILE_ONLY___ ═
489
+ 2024-04-15 13:25:21,243 INFO ___FILE_ONLY___ ═
490
+ 2024-04-15 13:25:21,243 INFO ___FILE_ONLY___ ═
491
+ 2024-04-15 13:25:21,243 INFO ___FILE_ONLY___ ═
492
+ 2024-04-15 13:25:21,243 INFO ___FILE_ONLY___ ═
493
+ 2024-04-15 13:25:21,243 INFO ___FILE_ONLY___ ═
494
+ 2024-04-15 13:25:21,243 INFO ___FILE_ONLY___ ═
495
+ 2024-04-15 13:25:21,244 INFO ___FILE_ONLY___ ═
496
+ 2024-04-15 13:25:21,244 INFO ___FILE_ONLY___ ═
497
+ 2024-04-15 13:25:21,244 INFO ___FILE_ONLY___ ═
498
+ 2024-04-15 13:25:21,244 INFO ___FILE_ONLY___ ═
499
+ 2024-04-15 13:25:21,244 INFO ___FILE_ONLY___ ═
500
+ 2024-04-15 13:25:21,278 INFO ___FILE_ONLY___ ═══════════════
501
+ 2024-04-15 13:25:21,279 INFO ___FILE_ONLY___ ═══════════════
502
+ 2024-04-15 13:25:21,279 INFO ___FILE_ONLY___ ╝
503
+
504
+ 2024-04-15 13:25:21,289 INFO ___FILE_ONLY___ ╔════════════════════════════════════════════════════════════╗
505
+
506
+ 2024-04-15 13:25:21,289 INFO ___FILE_ONLY___ ╠═ Installing: Google Cloud CRC32C Hash Tool ═╣
507
+
508
+ 2024-04-15 13:25:21,289 INFO ___FILE_ONLY___ ╚
509
+ 2024-04-15 13:25:21,294 INFO ___FILE_ONLY___ ════════════════════════════════════════════════════════════
510
+ 2024-04-15 13:25:21,294 INFO ___FILE_ONLY___ ╝
511
+
512
+ 2024-04-15 13:25:21,296 INFO ___FILE_ONLY___ ╔════════════════════════════════════════════════════════════╗
513
+
514
+ 2024-04-15 13:25:21,296 INFO ___FILE_ONLY___ ╠═ Installing: anthoscli ═╣
515
+
516
+ 2024-04-15 13:25:21,296 INFO ___FILE_ONLY___ ╚
517
+ 2024-04-15 13:25:21,300 DEBUG urllib3.connectionpool Starting new HTTPS connection (1): dl.google.com:443
518
+ 2024-04-15 13:25:21,360 DEBUG urllib3.connectionpool https://dl.google.com:443 "GET /dl/cloudsdk/channels/rapid/components/google-cloud-sdk-anthoscli-linux-x86_64-20240209195330.tar.gz HTTP/1.1" 200 72231225
519
+ 2024-04-15 13:25:21,643 INFO ___FILE_ONLY___ ═
520
+ 2024-04-15 13:25:21,646 INFO ___FILE_ONLY___ ═
521
+ 2024-04-15 13:25:21,649 INFO ___FILE_ONLY___ ═
522
+ 2024-04-15 13:25:21,653 INFO ___FILE_ONLY___ ═
523
+ 2024-04-15 13:25:21,656 INFO ___FILE_ONLY___ ═
524
+ 2024-04-15 13:25:21,659 INFO ___FILE_ONLY___ ═
525
+ 2024-04-15 13:25:21,662 INFO ___FILE_ONLY___ ═
526
+ 2024-04-15 13:25:21,665 INFO ___FILE_ONLY___ ═
527
+ 2024-04-15 13:25:21,668 INFO ___FILE_ONLY___ ═
528
+ 2024-04-15 13:25:21,671 INFO ___FILE_ONLY___ ═
529
+ 2024-04-15 13:25:21,675 INFO ___FILE_ONLY___ ═
530
+ 2024-04-15 13:25:21,678 INFO ___FILE_ONLY___ ═
531
+ 2024-04-15 13:25:21,681 INFO ___FILE_ONLY___ ═
532
+ 2024-04-15 13:25:21,684 INFO ___FILE_ONLY___ ═
533
+ 2024-04-15 13:25:21,687 INFO ___FILE_ONLY___ ═
534
+ 2024-04-15 13:25:21,690 INFO ___FILE_ONLY___ ═
535
+ 2024-04-15 13:25:21,693 INFO ___FILE_ONLY___ ═
536
+ 2024-04-15 13:25:21,696 INFO ___FILE_ONLY___ ═
537
+ 2024-04-15 13:25:21,699 INFO ___FILE_ONLY___ ═
538
+ 2024-04-15 13:25:21,703 INFO ___FILE_ONLY___ ═
539
+ 2024-04-15 13:25:21,706 INFO ___FILE_ONLY___ ═
540
+ 2024-04-15 13:25:21,709 INFO ___FILE_ONLY___ ═
541
+ 2024-04-15 13:25:21,712 INFO ___FILE_ONLY___ ═
542
+ 2024-04-15 13:25:21,715 INFO ___FILE_ONLY___ ═
543
+ 2024-04-15 13:25:21,718 INFO ___FILE_ONLY___ ═
544
+ 2024-04-15 13:25:21,721 INFO ___FILE_ONLY___ ═
545
+ 2024-04-15 13:25:21,725 INFO ___FILE_ONLY___ ═
546
+ 2024-04-15 13:25:21,728 INFO ___FILE_ONLY___ ═
547
+ 2024-04-15 13:25:21,731 INFO ___FILE_ONLY___ ═
548
+ 2024-04-15 13:25:21,735 INFO ___FILE_ONLY___ ═
549
+ 2024-04-15 13:25:23,940 INFO ___FILE_ONLY___ ══════════
550
+ 2024-04-15 13:25:23,945 INFO ___FILE_ONLY___ ═════════
551
+ 2024-04-15 13:25:23,971 INFO ___FILE_ONLY___ ═══════════
552
+ 2024-04-15 13:25:23,972 INFO ___FILE_ONLY___ ╝
553
+
554
+ 2024-04-15 13:25:23,994 INFO ___FILE_ONLY___ ╔═════════════════════════════════��══════════════════════════╗
555
+
556
+ 2024-04-15 13:25:23,994 INFO ___FILE_ONLY___ ╠═ Installing: anthoscli ═╣
557
+
558
+ 2024-04-15 13:25:23,994 INFO ___FILE_ONLY___ ╚
559
+ 2024-04-15 13:25:23,999 INFO ___FILE_ONLY___ ════════════════════════════════════════════════════════════
560
+ 2024-04-15 13:25:23,999 INFO ___FILE_ONLY___ ╝
561
+
562
+ 2024-04-15 13:25:24,001 INFO ___FILE_ONLY___ ╔════════════════════════════════════════════════════════════╗
563
+
564
+ 2024-04-15 13:25:24,001 INFO ___FILE_ONLY___ ╠═ Installing: gcloud cli dependencies ═╣
565
+
566
+ 2024-04-15 13:25:24,001 INFO ___FILE_ONLY___ ╚
567
+ 2024-04-15 13:25:24,005 DEBUG urllib3.connectionpool Starting new HTTPS connection (1): dl.google.com:443
568
+ 2024-04-15 13:25:24,067 DEBUG urllib3.connectionpool https://dl.google.com:443 "GET /dl/cloudsdk/channels/rapid/components/google-cloud-sdk-gcloud-deps-linux-x86_64-20210416153011.tar.gz HTTP/1.1" 200 104
569
+ 2024-04-15 13:25:24,067 INFO ___FILE_ONLY___ ══════════════════════════════
570
+ 2024-04-15 13:25:24,067 INFO ___FILE_ONLY___ ══════════════════════════════
571
+ 2024-04-15 13:25:24,068 INFO ___FILE_ONLY___ ╝
572
+
573
+ 2024-04-15 13:25:24,076 INFO ___FILE_ONLY___ ╔════════════════════════════════════════════════════════════╗
574
+
575
+ 2024-04-15 13:25:24,076 INFO ___FILE_ONLY___ ╠═ Creating backup and activating new installation ═╣
576
+
577
+ 2024-04-15 13:25:24,076 INFO ___FILE_ONLY___ ╚
578
+ 2024-04-15 13:25:24,076 DEBUG root Attempting to move directory [/tools/google-cloud-sdk] to [/tools/google-cloud-sdk.staging/.install/.backup]
579
+ 2024-04-15 13:25:24,076 INFO ___FILE_ONLY___ ══════════════════════════════
580
+ 2024-04-15 13:25:24,076 DEBUG root Attempting to move directory [/tools/google-cloud-sdk.staging] to [/tools/google-cloud-sdk]
581
+ 2024-04-15 13:25:24,076 INFO ___FILE_ONLY___ ══════════════════════════════
582
+ 2024-04-15 13:25:24,077 INFO ___FILE_ONLY___ ╝
583
+
584
+ 2024-04-15 13:25:24,080 DEBUG root Updating notification cache...
585
+ 2024-04-15 13:25:24,081 INFO ___FILE_ONLY___
586
+
587
+ 2024-04-15 13:25:24,083 INFO ___FILE_ONLY___ Performing post processing steps...
588
+ 2024-04-15 13:25:24,083 DEBUG root Executing command: ['/tools/google-cloud-sdk/bin/gcloud', 'components', 'post-process']
589
+ 2024-04-15 13:25:35,082 DEBUG ___FILE_ONLY___
590
+ 2024-04-15 13:25:35,082 DEBUG ___FILE_ONLY___
591
+ 2024-04-15 13:25:35,100 INFO ___FILE_ONLY___
592
+ Update done!
593
+
594
+
595
+ 2024-04-15 13:25:35,104 DEBUG root Chosen display Format:none
596
+ 2024-04-15 13:25:35,105 INFO root Display format: "none"
demo/.config/logs/2024.04.15/13.25.24.511550.log ADDED
@@ -0,0 +1,5 @@
 
 
 
 
 
 
1
+ 2024-04-15 13:25:24,512 DEBUG root Loaded Command Group: ['gcloud', 'components']
2
+ 2024-04-15 13:25:24,514 DEBUG root Loaded Command Group: ['gcloud', 'components', 'post_process']
3
+ 2024-04-15 13:25:24,516 DEBUG root Running [gcloud.components.post-process] with arguments: []
4
+ 2024-04-15 13:25:34,997 DEBUG root Chosen display Format:none
5
+ 2024-04-15 13:25:34,997 INFO root Display format: "none"
demo/.config/logs/2024.04.15/13.25.35.658011.log ADDED
@@ -0,0 +1,169 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ 2024-04-15 13:25:35,659 DEBUG root Loaded Command Group: ['gcloud', 'components']
2
+ 2024-04-15 13:25:35,661 DEBUG root Loaded Command Group: ['gcloud', 'components', 'update']
3
+ 2024-04-15 13:25:35,664 DEBUG root Running [gcloud.components.update] with arguments: [--quiet: "True", COMPONENT-IDS:8: "['gcloud', 'core', 'bq', 'gsutil', 'compute', 'preview', 'alpha', 'beta']"]
4
+ 2024-04-15 13:25:35,665 INFO ___FILE_ONLY___ Beginning update. This process may take several minutes.
5
+
6
+ 2024-04-15 13:25:35,673 DEBUG urllib3.connectionpool Starting new HTTPS connection (1): dl.google.com:443
7
+ 2024-04-15 13:25:35,731 DEBUG urllib3.connectionpool https://dl.google.com:443 "GET /dl/cloudsdk/channels/rapid/components-2.json HTTP/1.1" 200 214439
8
+ 2024-04-15 13:25:35,751 WARNING root Component [compute] no longer exists.
9
+ 2024-04-15 13:25:35,751 WARNING root Component [preview] no longer exists.
10
+ 2024-04-15 13:25:35,752 INFO ___FILE_ONLY___
11
+
12
+ 2024-04-15 13:25:35,753 INFO ___FILE_ONLY___
13
+ Your current Google Cloud CLI version is: 471.0.0
14
+
15
+ 2024-04-15 13:25:35,753 INFO ___FILE_ONLY___ Installing components from version: 471.0.0
16
+
17
+ 2024-04-15 13:25:35,753 INFO ___FILE_ONLY___
18
+
19
+ 2024-04-15 13:25:35,753 DEBUG root Chosen display Format:table[box,title="These components will be removed."](details.display_name:label=Name:align=left,version.version_string:label=Version:align=right,data.size.size(zero="",min=1048576):label=Size:align=right)
20
+ 2024-04-15 13:25:35,754 DEBUG root Chosen display Format:table[box,title="These components will be updated."](details.display_name:label=Name:align=left,version.version_string:label=Version:align=right,data.size.size(zero="",min=1048576):label=Size:align=right)
21
+ 2024-04-15 13:25:35,755 DEBUG root Chosen display Format:table[box,title="These components will be installed."](details.display_name:label=Name:align=left,version.version_string:label=Version:align=right,data.size.size(zero="",min=1048576):label=Size:align=right)
22
+ 2024-04-15 13:25:35,793 INFO ___FILE_ONLY___ ┌──────────────────────────────────────────────┐
23
+ 2024-04-15 13:25:35,794 INFO ___FILE_ONLY___
24
+
25
+ 2024-04-15 13:25:35,794 INFO ___FILE_ONLY___ │ These components will be installed. │
26
+ 2024-04-15 13:25:35,794 INFO ___FILE_ONLY___
27
+
28
+ 2024-04-15 13:25:35,794 INFO ___FILE_ONLY___ ├───────────────────────┬────────────┬─────────┤
29
+ 2024-04-15 13:25:35,794 INFO ___FILE_ONLY___
30
+
31
+ 2024-04-15 13:25:35,794 INFO ___FILE_ONLY___ │ Name │ Version │ Size │
32
+ 2024-04-15 13:25:35,794 INFO ___FILE_ONLY___
33
+
34
+ 2024-04-15 13:25:35,794 INFO ___FILE_ONLY___ ├───────────────────────┼────────────┼─────────┤
35
+ 2024-04-15 13:25:35,794 INFO ___FILE_ONLY___
36
+
37
+ 2024-04-15 13:25:35,794 INFO ___FILE_ONLY___ │
38
+ 2024-04-15 13:25:35,794 INFO ___FILE_ONLY___ gcloud Alpha Commands
39
+ 2024-04-15 13:25:35,794 INFO ___FILE_ONLY___
40
+ 2024-04-15 13:25:35,795 INFO ___FILE_ONLY___ │
41
+ 2024-04-15 13:25:35,795 INFO ___FILE_ONLY___ 2024.03.29
42
+ 2024-04-15 13:25:35,795 INFO ___FILE_ONLY___
43
+ 2024-04-15 13:25:35,795 INFO ___FILE_ONLY___ │
44
+ 2024-04-15 13:25:35,795 INFO ___FILE_ONLY___ < 1 MiB
45
+ 2024-04-15 13:25:35,795 INFO ___FILE_ONLY___
46
+ 2024-04-15 13:25:35,795 INFO ___FILE_ONLY___ │
47
+ 2024-04-15 13:25:35,795 INFO ___FILE_ONLY___
48
+
49
+ 2024-04-15 13:25:35,795 INFO ___FILE_ONLY___ │
50
+ 2024-04-15 13:25:35,795 INFO ___FILE_ONLY___ gcloud Beta Commands
51
+ 2024-04-15 13:25:35,795 INFO ___FILE_ONLY___
52
+ 2024-04-15 13:25:35,795 INFO ___FILE_ONLY___ │
53
+ 2024-04-15 13:25:35,795 INFO ___FILE_ONLY___ 2024.03.29
54
+ 2024-04-15 13:25:35,795 INFO ___FILE_ONLY___
55
+ 2024-04-15 13:25:35,796 INFO ___FILE_ONLY___ │
56
+ 2024-04-15 13:25:35,796 INFO ___FILE_ONLY___ < 1 MiB
57
+ 2024-04-15 13:25:35,796 INFO ___FILE_ONLY___
58
+ 2024-04-15 13:25:35,796 INFO ___FILE_ONLY___ │
59
+ 2024-04-15 13:25:35,796 INFO ___FILE_ONLY___
60
+
61
+ 2024-04-15 13:25:35,796 INFO ___FILE_ONLY___ └───────────────────────┴────────────┴─────────┘
62
+ 2024-04-15 13:25:35,796 INFO ___FILE_ONLY___
63
+
64
+ 2024-04-15 13:25:35,796 INFO ___FILE_ONLY___
65
+
66
+ 2024-04-15 13:25:35,800 DEBUG urllib3.connectionpool Starting new HTTPS connection (1): dl.google.com:443
67
+ 2024-04-15 13:25:35,860 DEBUG urllib3.connectionpool https://dl.google.com:443 "GET /dl/cloudsdk/channels/rapid/RELEASE_NOTES HTTP/1.1" 200 1186568
68
+ 2024-04-15 13:25:35,931 INFO ___FILE_ONLY___ For the latest full release notes, please visit:
69
+ https://cloud.google.com/sdk/release_notes
70
+
71
+
72
+ 2024-04-15 13:25:35,934 INFO ___FILE_ONLY___ ╔════════════════════════════════════════════════════════════╗
73
+
74
+ 2024-04-15 13:25:35,934 INFO ___FILE_ONLY___ ╠═ Creating update staging area ═╣
75
+
76
+ 2024-04-15 13:25:35,934 INFO ___FILE_ONLY___ ╚
77
+ 2024-04-15 13:25:35,934 INFO ___FILE_ONLY___ ══════
78
+ 2024-04-15 13:25:36,701 INFO ___FILE_ONLY___ ══════
79
+ 2024-04-15 13:25:36,701 INFO ___FILE_ONLY___ ══════
80
+ 2024-04-15 13:25:37,165 INFO ___FILE_ONLY___ ═
81
+ 2024-04-15 13:25:37,472 INFO ___FILE_ONLY___ ═
82
+ 2024-04-15 13:25:37,530 INFO ___FILE_ONLY___ ═
83
+ 2024-04-15 13:25:37,592 INFO ___FILE_ONLY___ ═
84
+ 2024-04-15 13:25:37,639 INFO ___FILE_ONLY___ ═
85
+ 2024-04-15 13:25:37,684 INFO ___FILE_ONLY___ ═
86
+ 2024-04-15 13:25:37,727 INFO ___FILE_ONLY___ ═
87
+ 2024-04-15 13:25:37,774 INFO ___FILE_ONLY___ ═
88
+ 2024-04-15 13:25:37,853 INFO ___FILE_ONLY___ ═
89
+ 2024-04-15 13:25:37,978 INFO ___FILE_ONLY___ ═
90
+ 2024-04-15 13:25:38,148 INFO ___FILE_ONLY___ ═
91
+ 2024-04-15 13:25:38,241 INFO ___FILE_ONLY___ ═
92
+ 2024-04-15 13:25:38,329 INFO ___FILE_ONLY___ ═
93
+ 2024-04-15 13:25:38,415 INFO ___FILE_ONLY___ ═
94
+ 2024-04-15 13:25:38,494 INFO ___FILE_ONLY___ ═
95
+ 2024-04-15 13:25:38,567 INFO ___FILE_ONLY___ ═
96
+ 2024-04-15 13:25:38,679 INFO ___FILE_ONLY___ ═
97
+ 2024-04-15 13:25:38,750 INFO ___FILE_ONLY___ ═
98
+ 2024-04-15 13:25:38,808 INFO ___FILE_ONLY___ ═
99
+ 2024-04-15 13:25:38,873 INFO ___FILE_ONLY___ ═
100
+ 2024-04-15 13:25:38,940 INFO ___FILE_ONLY___ ═
101
+ 2024-04-15 13:25:38,998 INFO ___FILE_ONLY___ ═
102
+ 2024-04-15 13:25:39,065 INFO ___FILE_ONLY___ ═
103
+ 2024-04-15 13:25:39,133 INFO ___FILE_ONLY___ ═
104
+ 2024-04-15 13:25:39,202 INFO ___FILE_ONLY___ ═
105
+ 2024-04-15 13:25:39,263 INFO ___FILE_ONLY___ ═
106
+ 2024-04-15 13:25:39,325 INFO ___FILE_ONLY___ ═
107
+ 2024-04-15 13:25:39,407 INFO ___FILE_ONLY___ ═
108
+ 2024-04-15 13:25:39,484 INFO ___FILE_ONLY___ ═
109
+ 2024-04-15 13:25:39,557 INFO ___FILE_ONLY___ ═
110
+ 2024-04-15 13:25:39,641 INFO ___FILE_ONLY___ ═
111
+ 2024-04-15 13:25:39,797 INFO ___FILE_ONLY___ ═
112
+ 2024-04-15 13:25:39,875 INFO ___FILE_ONLY___ ═
113
+ 2024-04-15 13:25:39,969 INFO ___FILE_ONLY___ ═
114
+ 2024-04-15 13:25:40,053 INFO ___FILE_ONLY___ ═
115
+ 2024-04-15 13:25:40,127 INFO ___FILE_ONLY___ ═
116
+ 2024-04-15 13:25:40,193 INFO ___FILE_ONLY___ ═
117
+ 2024-04-15 13:25:40,282 INFO ___FILE_ONLY___ ═
118
+ 2024-04-15 13:25:40,342 INFO ___FILE_ONLY___ ═
119
+ 2024-04-15 13:25:40,420 INFO ___FILE_ONLY___ ═
120
+ 2024-04-15 13:25:40,481 INFO ___FILE_ONLY___ ═
121
+ 2024-04-15 13:25:40,548 INFO ___FILE_ONLY___ ═
122
+ 2024-04-15 13:25:40,549 INFO ___FILE_ONLY___ ╝
123
+
124
+ 2024-04-15 13:25:44,558 INFO ___FILE_ONLY___ ╔════════════════════════════════════════════════════════════╗
125
+
126
+ 2024-04-15 13:25:44,559 INFO ___FILE_ONLY___ ╠═ Installing: gcloud Alpha Commands ═╣
127
+
128
+ 2024-04-15 13:25:44,559 INFO ___FILE_ONLY___ ╚
129
+ 2024-04-15 13:25:44,563 DEBUG urllib3.connectionpool Starting new HTTPS connection (1): dl.google.com:443
130
+ 2024-04-15 13:25:44,626 DEBUG urllib3.connectionpool https://dl.google.com:443 "GET /dl/cloudsdk/channels/rapid/components/google-cloud-sdk-alpha-20240329151455.tar.gz HTTP/1.1" 200 800
131
+ 2024-04-15 13:25:44,626 INFO ___FILE_ONLY___ ══════════════════════════════
132
+ 2024-04-15 13:25:44,628 INFO ___FILE_ONLY___ ══════════════════════════════
133
+ 2024-04-15 13:25:44,628 INFO ___FILE_ONLY___ ╝
134
+
135
+ 2024-04-15 13:25:44,637 INFO ___FILE_ONLY___ ╔════════════════════════════════════════════════════════════╗
136
+
137
+ 2024-04-15 13:25:44,637 INFO ___FILE_ONLY___ ╠═ Installing: gcloud Beta Commands ═╣
138
+
139
+ 2024-04-15 13:25:44,637 INFO ___FILE_ONLY___ ╚
140
+ 2024-04-15 13:25:44,641 DEBUG urllib3.connectionpool Starting new HTTPS connection (1): dl.google.com:443
141
+ 2024-04-15 13:25:44,704 DEBUG urllib3.connectionpool https://dl.google.com:443 "GET /dl/cloudsdk/channels/rapid/components/google-cloud-sdk-beta-20240329151455.tar.gz HTTP/1.1" 200 797
142
+ 2024-04-15 13:25:44,705 INFO ___FILE_ONLY___ ══════════════════════════════
143
+ 2024-04-15 13:25:44,706 INFO ___FILE_ONLY___ ══════════════════════════════
144
+ 2024-04-15 13:25:44,706 INFO ___FILE_ONLY___ ╝
145
+
146
+ 2024-04-15 13:25:44,716 INFO ___FILE_ONLY___ ��════════════════════════════════════════════════════════════╗
147
+
148
+ 2024-04-15 13:25:44,716 INFO ___FILE_ONLY___ ╠═ Creating backup and activating new installation ═╣
149
+
150
+ 2024-04-15 13:25:44,716 INFO ___FILE_ONLY___ ╚
151
+ 2024-04-15 13:25:44,717 DEBUG root Attempting to move directory [/tools/google-cloud-sdk] to [/tools/google-cloud-sdk.staging/.install/.backup]
152
+ 2024-04-15 13:25:44,717 INFO ___FILE_ONLY___ ══════════════════════════════
153
+ 2024-04-15 13:25:44,717 DEBUG root Attempting to move directory [/tools/google-cloud-sdk.staging] to [/tools/google-cloud-sdk]
154
+ 2024-04-15 13:25:44,717 INFO ___FILE_ONLY___ ══════════════════════════════
155
+ 2024-04-15 13:25:44,717 INFO ___FILE_ONLY___ ╝
156
+
157
+ 2024-04-15 13:25:44,722 DEBUG root Updating notification cache...
158
+ 2024-04-15 13:25:44,722 INFO ___FILE_ONLY___
159
+
160
+ 2024-04-15 13:25:44,724 INFO ___FILE_ONLY___ Performing post processing steps...
161
+ 2024-04-15 13:25:44,725 DEBUG root Executing command: ['/tools/google-cloud-sdk/bin/gcloud', 'components', 'post-process']
162
+ 2024-04-15 13:25:59,176 DEBUG ___FILE_ONLY___
163
+ 2024-04-15 13:25:59,177 DEBUG ___FILE_ONLY___
164
+ 2024-04-15 13:25:59,247 INFO ___FILE_ONLY___
165
+ Update done!
166
+
167
+
168
+ 2024-04-15 13:25:59,251 DEBUG root Chosen display Format:none
169
+ 2024-04-15 13:25:59,252 INFO root Display format: "none"
demo/.config/logs/2024.04.15/13.25.45.199675.log ADDED
@@ -0,0 +1,5 @@
 
 
 
 
 
 
1
+ 2024-04-15 13:25:45,200 DEBUG root Loaded Command Group: ['gcloud', 'components']
2
+ 2024-04-15 13:25:45,202 DEBUG root Loaded Command Group: ['gcloud', 'components', 'post_process']
3
+ 2024-04-15 13:25:45,205 DEBUG root Running [gcloud.components.post-process] with arguments: []
4
+ 2024-04-15 13:25:59,091 DEBUG root Chosen display Format:none
5
+ 2024-04-15 13:25:59,092 INFO root Display format: "none"
demo/.config/logs/2024.04.15/13.25.59.817323.log ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ 2024-04-15 13:25:59,819 DEBUG root Loaded Command Group: ['gcloud', 'config']
2
+ 2024-04-15 13:25:59,944 DEBUG root Loaded Command Group: ['gcloud', 'config', 'set']
3
+ 2024-04-15 13:25:59,947 DEBUG root Running [gcloud.config.set] with arguments: [SECTION/PROPERTY: "component_manager/disable_update_check", VALUE: "true"]
4
+ 2024-04-15 13:25:59,948 INFO ___FILE_ONLY___ Updated property [component_manager/disable_update_check].
5
+
6
+ 2024-04-15 13:25:59,949 DEBUG root Chosen display Format:default
7
+ 2024-04-15 13:25:59,950 INFO root Display format: "default"
8
+ 2024-04-15 13:25:59,950 DEBUG root SDK update checks are disabled.
demo/.config/logs/2024.04.15/13.26.00.519914.log ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ 2024-04-15 13:26:00,522 DEBUG root Loaded Command Group: ['gcloud', 'config']
2
+ 2024-04-15 13:26:00,649 DEBUG root Loaded Command Group: ['gcloud', 'config', 'set']
3
+ 2024-04-15 13:26:00,652 DEBUG root Running [gcloud.config.set] with arguments: [SECTION/PROPERTY: "compute/gce_metadata_read_timeout_sec", VALUE: "0"]
4
+ 2024-04-15 13:26:00,653 INFO ___FILE_ONLY___ Updated property [compute/gce_metadata_read_timeout_sec].
5
+
6
+ 2024-04-15 13:26:00,654 DEBUG root Chosen display Format:default
7
+ 2024-04-15 13:26:00,655 INFO root Display format: "default"
8
+ 2024-04-15 13:26:00,655 DEBUG root SDK update checks are disabled.
demo/.gitattributes ADDED
@@ -0,0 +1,38 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ *.7z filter=lfs diff=lfs merge=lfs -text
2
+ *.arrow filter=lfs diff=lfs merge=lfs -text
3
+ *.bin filter=lfs diff=lfs merge=lfs -text
4
+ *.bz2 filter=lfs diff=lfs merge=lfs -text
5
+ *.ckpt filter=lfs diff=lfs merge=lfs -text
6
+ *.ftz filter=lfs diff=lfs merge=lfs -text
7
+ *.gz filter=lfs diff=lfs merge=lfs -text
8
+ *.h5 filter=lfs diff=lfs merge=lfs -text
9
+ *.joblib filter=lfs diff=lfs merge=lfs -text
10
+ *.lfs.* filter=lfs diff=lfs merge=lfs -text
11
+ *.mlmodel filter=lfs diff=lfs merge=lfs -text
12
+ *.model filter=lfs diff=lfs merge=lfs -text
13
+ *.msgpack filter=lfs diff=lfs merge=lfs -text
14
+ *.npy filter=lfs diff=lfs merge=lfs -text
15
+ *.npz filter=lfs diff=lfs merge=lfs -text
16
+ *.onnx filter=lfs diff=lfs merge=lfs -text
17
+ *.ot filter=lfs diff=lfs merge=lfs -text
18
+ *.parquet filter=lfs diff=lfs merge=lfs -text
19
+ *.pb filter=lfs diff=lfs merge=lfs -text
20
+ *.pickle filter=lfs diff=lfs merge=lfs -text
21
+ *.pkl filter=lfs diff=lfs merge=lfs -text
22
+ *.pt filter=lfs diff=lfs merge=lfs -text
23
+ *.pth filter=lfs diff=lfs merge=lfs -text
24
+ *.rar filter=lfs diff=lfs merge=lfs -text
25
+ *.safetensors filter=lfs diff=lfs merge=lfs -text
26
+ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
27
+ *.tar.* filter=lfs diff=lfs merge=lfs -text
28
+ *.tar filter=lfs diff=lfs merge=lfs -text
29
+ *.tflite filter=lfs diff=lfs merge=lfs -text
30
+ *.tgz filter=lfs diff=lfs merge=lfs -text
31
+ *.wasm filter=lfs diff=lfs merge=lfs -text
32
+ *.xz filter=lfs diff=lfs merge=lfs -text
33
+ *.zip filter=lfs diff=lfs merge=lfs -text
34
+ *.zst filter=lfs diff=lfs merge=lfs -text
35
+ *tfevents* filter=lfs diff=lfs merge=lfs -text
36
+ dust3r/checkpoints/DUSt3R_ViTLarge_BaseDecoder_512_dpt.pth.1 filter=lfs diff=lfs merge=lfs -text
37
+ sample_data/mnist_test.csv filter=lfs diff=lfs merge=lfs -text
38
+ sample_data/mnist_train_small.csv filter=lfs diff=lfs merge=lfs -text
demo/README.md ADDED
@@ -0,0 +1,6 @@
 
 
 
 
 
 
 
1
+ ---
2
+ title: demo
3
+ app_file: /content/dust3r/demo.py
4
+ sdk: gradio
5
+ sdk_version: 4.26.0
6
+ ---
demo/app.py ADDED
@@ -0,0 +1,7 @@
 
 
 
 
 
 
 
 
1
+ import gradio as gr
2
+
3
+ def greet(name):
4
+ return "Hello " + name + "!!"
5
+
6
+ iface = gr.Interface(fn=greet, inputs="text", outputs="text")
7
+ iface.launch()
demo/dust3r/.gitignore ADDED
@@ -0,0 +1,132 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ data/
2
+ checkpoints/
3
+
4
+ # Byte-compiled / optimized / DLL files
5
+ __pycache__/
6
+ *.py[cod]
7
+ *$py.class
8
+
9
+ # C extensions
10
+ *.so
11
+
12
+ # Distribution / packaging
13
+ .Python
14
+ build/
15
+ develop-eggs/
16
+ dist/
17
+ downloads/
18
+ eggs/
19
+ .eggs/
20
+ lib/
21
+ lib64/
22
+ parts/
23
+ sdist/
24
+ var/
25
+ wheels/
26
+ pip-wheel-metadata/
27
+ share/python-wheels/
28
+ *.egg-info/
29
+ .installed.cfg
30
+ *.egg
31
+ MANIFEST
32
+
33
+ # PyInstaller
34
+ # Usually these files are written by a python script from a template
35
+ # before PyInstaller builds the exe, so as to inject date/other infos into it.
36
+ *.manifest
37
+ *.spec
38
+
39
+ # Installer logs
40
+ pip-log.txt
41
+ pip-delete-this-directory.txt
42
+
43
+ # Unit test / coverage reports
44
+ htmlcov/
45
+ .tox/
46
+ .nox/
47
+ .coverage
48
+ .coverage.*
49
+ .cache
50
+ nosetests.xml
51
+ coverage.xml
52
+ *.cover
53
+ *.py,cover
54
+ .hypothesis/
55
+ .pytest_cache/
56
+
57
+ # Translations
58
+ *.mo
59
+ *.pot
60
+
61
+ # Django stuff:
62
+ *.log
63
+ local_settings.py
64
+ db.sqlite3
65
+ db.sqlite3-journal
66
+
67
+ # Flask stuff:
68
+ instance/
69
+ .webassets-cache
70
+
71
+ # Scrapy stuff:
72
+ .scrapy
73
+
74
+ # Sphinx documentation
75
+ docs/_build/
76
+
77
+ # PyBuilder
78
+ target/
79
+
80
+ # Jupyter Notebook
81
+ .ipynb_checkpoints
82
+
83
+ # IPython
84
+ profile_default/
85
+ ipython_config.py
86
+
87
+ # pyenv
88
+ .python-version
89
+
90
+ # pipenv
91
+ # According to pypa/pipenv#598, it is recommended to include Pipfile.lock in version control.
92
+ # However, in case of collaboration, if having platform-specific dependencies or dependencies
93
+ # having no cross-platform support, pipenv may install dependencies that don't work, or not
94
+ # install all needed dependencies.
95
+ #Pipfile.lock
96
+
97
+ # PEP 582; used by e.g. github.com/David-OConnor/pyflow
98
+ __pypackages__/
99
+
100
+ # Celery stuff
101
+ celerybeat-schedule
102
+ celerybeat.pid
103
+
104
+ # SageMath parsed files
105
+ *.sage.py
106
+
107
+ # Environments
108
+ .env
109
+ .venv
110
+ env/
111
+ venv/
112
+ ENV/
113
+ env.bak/
114
+ venv.bak/
115
+
116
+ # Spyder project settings
117
+ .spyderproject
118
+ .spyproject
119
+
120
+ # Rope project settings
121
+ .ropeproject
122
+
123
+ # mkdocs documentation
124
+ /site
125
+
126
+ # mypy
127
+ .mypy_cache/
128
+ .dmypy.json
129
+ dmypy.json
130
+
131
+ # Pyre type checker
132
+ .pyre/
demo/dust3r/.gitmodules ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ [submodule "croco"]
2
+ path = croco
3
+ url = https://github.com/naver/croco
demo/dust3r/LICENSE ADDED
@@ -0,0 +1,7 @@
 
 
 
 
 
 
 
 
1
+ DUSt3R, Copyright (c) 2024-present Naver Corporation, is licensed under the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 license.
2
+
3
+ A summary of the CC BY-NC-SA 4.0 license is located here:
4
+ https://creativecommons.org/licenses/by-nc-sa/4.0/
5
+
6
+ The CC BY-NC-SA 4.0 license is located here:
7
+ https://creativecommons.org/licenses/by-nc-sa/4.0/legalcode
demo/dust3r/NOTICE ADDED
@@ -0,0 +1,13 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ DUSt3R
2
+ Copyright 2024-present NAVER Corp.
3
+
4
+ This project contains subcomponents with separate copyright notices and license terms.
5
+ Your use of the source code for these subcomponents is subject to the terms and conditions of the following licenses.
6
+
7
+ ====
8
+
9
+ naver/croco
10
+ https://github.com/naver/croco/
11
+
12
+ Creative Commons Attribution-NonCommercial-ShareAlike 4.0
13
+
demo/dust3r/README.md ADDED
@@ -0,0 +1,299 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # DUSt3R
2
+
3
+ Official implementation of `DUSt3R: Geometric 3D Vision Made Easy`
4
+ [[Project page](https://dust3r.europe.naverlabs.com/)], [[DUSt3R arxiv](https://arxiv.org/abs/2312.14132)]
5
+
6
+ ![Example of reconstruction from two images](assets/pipeline1.jpg)
7
+
8
+ ![High level overview of DUSt3R capabilities](assets/dust3r_archi.jpg)
9
+
10
+ ```bibtex
11
+ @misc{wang2023dust3r,
12
+ title={DUSt3R: Geometric 3D Vision Made Easy},
13
+ author={Shuzhe Wang and Vincent Leroy and Yohann Cabon and Boris Chidlovskii and Jerome Revaud},
14
+ year={2023},
15
+ eprint={2312.14132},
16
+ archivePrefix={arXiv},
17
+ primaryClass={cs.CV}
18
+ }
19
+ ```
20
+
21
+ ## Table of Contents
22
+ - [DUSt3R](#dust3r)
23
+ - [License](#license)
24
+ - [Get Started](#get-started)
25
+ - [Installation](#installation)
26
+ - [Checkpoints](#checkpoints)
27
+ - [Interactive demo](#interactive-demo)
28
+ - [Usage](#usage)
29
+ - [Training](#training)
30
+ - [Demo](#demo)
31
+ - [Our Hyperparameters](#our-hyperparameters)
32
+
33
+ ## License
34
+ The code is distributed under the CC BY-NC-SA 4.0 License. See See [LICENSE](LICENSE) for more information.
35
+ ```python
36
+ # Copyright (C) 2024-present Naver Corporation. All rights reserved.
37
+ # Licensed under CC BY-NC-SA 4.0 (non-commercial use only).
38
+ ```
39
+
40
+ ## Get Started
41
+
42
+ ### Installation
43
+
44
+ 1. Clone DUSt3R
45
+ ```bash
46
+ git clone --recursive https://github.com/naver/dust3r
47
+ cd dust3r
48
+ # if you have already cloned dust3r:
49
+ # git submodule update --init --recursive
50
+ ```
51
+
52
+ 2. Create the environment, here we show an example using conda.
53
+ ```bash
54
+ conda create -n dust3r python=3.11 cmake=3.14.0
55
+ conda activate dust3r
56
+ conda install pytorch torchvision pytorch-cuda=12.1 -c pytorch -c nvidia # use the correct version of cuda for your system
57
+ pip install -r requirements.txt
58
+ ```
59
+
60
+
61
+ 3. Optional, compile the cuda kernels for RoPE (as in CroCo v2)
62
+ ```bash
63
+ # DUST3R relies on RoPE positional embeddings for which you can compile some cuda kernels for faster runtime.
64
+ cd croco/models/curope/
65
+ python setup.py build_ext --inplace
66
+ cd ../../../
67
+ ```
68
+
69
+ 4. Download pre-trained model
70
+ ```bash
71
+ mkdir -p checkpoints/
72
+ wget https://download.europe.naverlabs.com/ComputerVision/DUSt3R/DUSt3R_ViTLarge_BaseDecoder_512_dpt.pth -P checkpoints/
73
+ ```
74
+
75
+ ### Checkpoints
76
+
77
+ We provide several pre-trained models:
78
+
79
+ | Modelname | Training resolutions | Head | Encoder | Decoder |
80
+ |-------------|----------------------|------|---------|---------|
81
+ | [`DUSt3R_ViTLarge_BaseDecoder_224_linear.pth`](https://download.europe.naverlabs.com/ComputerVision/DUSt3R/DUSt3R_ViTLarge_BaseDecoder_224_linear.pth) | 224x224 | Linear | ViT-L | ViT-B |
82
+ | [`DUSt3R_ViTLarge_BaseDecoder_512_linear.pth`](https://download.europe.naverlabs.com/ComputerVision/DUSt3R/DUSt3R_ViTLarge_BaseDecoder_512_linear.pth) | 512x384, 512x336, 512x288, 512x256, 512x160 | Linear | ViT-L | ViT-B |
83
+ | [`DUSt3R_ViTLarge_BaseDecoder_512_dpt.pth`](https://download.europe.naverlabs.com/ComputerVision/DUSt3R/DUSt3R_ViTLarge_BaseDecoder_512_dpt.pth) | 512x384, 512x336, 512x288, 512x256, 512x160 | DPT | ViT-L | ViT-B |
84
+
85
+ You can check the hyperparameters we used to train these models in the [section: Our Hyperparameters](#our-hyperparameters)
86
+
87
+ ### Interactive demo
88
+ In this demo, you should be able run DUSt3R on your machine to reconstruct a scene.
89
+ First select images that depicts the same scene.
90
+
91
+ You can ajust the global alignment schedule and its number of iterations.
92
+ Note: if you selected one or two images, the global alignment procedure will be skipped (mode=GlobalAlignerMode.PairViewer)
93
+ Hit "Run" and wait.
94
+ When the global alignment ends, the reconstruction appears.
95
+ Use the slider "min_conf_thr" to show or remove low confidence areas.
96
+
97
+ ```bash
98
+ python3 demo.py --weights checkpoints/DUSt3R_ViTLarge_BaseDecoder_512_dpt.pth
99
+
100
+ # Use --image_size to select the correct resolution for your checkpoint. 512 (default) or 224
101
+ # Use --local_network to make it accessible on the local network, or --server_name to specify the url manually
102
+ # Use --server_port to change the port, by default it will search for an available port starting at 7860
103
+ # Use --device to use a different device, by default it's "cuda"
104
+ ```
105
+
106
+ ![demo](assets/demo.jpg)
107
+
108
+ ## Usage
109
+
110
+ ```python
111
+ from dust3r.inference import inference, load_model
112
+ from dust3r.utils.image import load_images
113
+ from dust3r.image_pairs import make_pairs
114
+ from dust3r.cloud_opt import global_aligner, GlobalAlignerMode
115
+
116
+ if __name__ == '__main__':
117
+ model_path = "checkpoints/DUSt3R_ViTLarge_BaseDecoder_512_dpt.pth"
118
+ device = 'cuda'
119
+ batch_size = 1
120
+ schedule = 'cosine'
121
+ lr = 0.01
122
+ niter = 300
123
+
124
+ model = load_model(model_path, device)
125
+ # load_images can take a list of images or a directory
126
+ images = load_images(['croco/assets/Chateau1.png', 'croco/assets/Chateau2.png'], size=512)
127
+ pairs = make_pairs(images, scene_graph='complete', prefilter=None, symmetrize=True)
128
+ output = inference(pairs, model, device, batch_size=batch_size)
129
+
130
+ # at this stage, you have the raw dust3r predictions
131
+ view1, pred1 = output['view1'], output['pred1']
132
+ view2, pred2 = output['view2'], output['pred2']
133
+ # here, view1, pred1, view2, pred2 are dicts of lists of len(2)
134
+ # -> because we symmetrize we have (im1, im2) and (im2, im1) pairs
135
+ # in each view you have:
136
+ # an integer image identifier: view1['idx'] and view2['idx']
137
+ # the img: view1['img'] and view2['img']
138
+ # the image shape: view1['true_shape'] and view2['true_shape']
139
+ # an instance string output by the dataloader: view1['instance'] and view2['instance']
140
+ # pred1 and pred2 contains the confidence values: pred1['conf'] and pred2['conf']
141
+ # pred1 contains 3D points for view1['img'] in view1['img'] space: pred1['pts3d']
142
+ # pred2 contains 3D points for view2['img'] in view1['img'] space: pred2['pts3d_in_other_view']
143
+
144
+ # next we'll use the global_aligner to align the predictions
145
+ # depending on your task, you may be fine with the raw output and not need it
146
+ # with only two input images, you could use GlobalAlignerMode.PairViewer: it would just convert the output
147
+ # if using GlobalAlignerMode.PairViewer, no need to run compute_global_alignment
148
+ scene = global_aligner(output, device=device, mode=GlobalAlignerMode.PointCloudOptimizer)
149
+ loss = scene.compute_global_alignment(init="mst", niter=niter, schedule=schedule, lr=lr)
150
+
151
+ # retrieve useful values from scene:
152
+ imgs = scene.imgs
153
+ focals = scene.get_focals()
154
+ poses = scene.get_im_poses()
155
+ pts3d = scene.get_pts3d()
156
+ confidence_masks = scene.get_masks()
157
+
158
+ # visualize reconstruction
159
+ scene.show()
160
+
161
+ # find 2D-2D matches between the two images
162
+ from dust3r.utils.geometry import find_reciprocal_matches, xy_grid
163
+ pts2d_list, pts3d_list = [], []
164
+ for i in range(2):
165
+ conf_i = confidence_masks[i].cpu().numpy()
166
+ pts2d_list.append(xy_grid(*imgs[i].shape[:2][::-1])[conf_i]) # imgs[i].shape[:2] = (H, W)
167
+ pts3d_list.append(pts3d[i].detach().cpu().numpy()[conf_i])
168
+ reciprocal_in_P2, nn2_in_P1, num_matches = find_reciprocal_matches(*pts3d_list)
169
+ print(f'found {num_matches} matches')
170
+ matches_im1 = pts2d_list[1][reciprocal_in_P2]
171
+ matches_im0 = pts2d_list[0][nn2_in_P1][reciprocal_in_P2]
172
+
173
+ # visualize a few matches
174
+ import numpy as np
175
+ from matplotlib import pyplot as pl
176
+ n_viz = 10
177
+ match_idx_to_viz = np.round(np.linspace(0, num_matches-1, n_viz)).astype(int)
178
+ viz_matches_im0, viz_matches_im1 = matches_im0[match_idx_to_viz], matches_im1[match_idx_to_viz]
179
+
180
+ H0, W0, H1, W1 = *imgs[0].shape[:2], *imgs[1].shape[:2]
181
+ img0 = np.pad(imgs[0], ((0, max(H1 - H0, 0)), (0, 0), (0, 0)), 'constant', constant_values=0)
182
+ img1 = np.pad(imgs[1], ((0, max(H0 - H1, 0)), (0, 0), (0, 0)), 'constant', constant_values=0)
183
+ img = np.concatenate((img0, img1), axis=1)
184
+ pl.figure()
185
+ pl.imshow(img)
186
+ cmap = pl.get_cmap('jet')
187
+ for i in range(n_viz):
188
+ (x0, y0), (x1, y1) = viz_matches_im0[i].T, viz_matches_im1[i].T
189
+ pl.plot([x0, x1 + W0], [y0, y1], '-+', color=cmap(i / (n_viz - 1)), scalex=False, scaley=False)
190
+ pl.show(block=True)
191
+
192
+ ```
193
+ ![matching example on croco pair](assets/matching.jpg)
194
+
195
+ ## Training
196
+ In this section, we present propose a short demonstration to get started with training DUSt3R. At the moment, we didn't release the training datasets, so we're going to download and prepare a subset of [CO3Dv2](https://github.com/facebookresearch/co3d) - [Creative Commons Attribution-NonCommercial 4.0 International](https://github.com/facebookresearch/co3d/blob/main/LICENSE) and launch the training code on it.
197
+ The demo model will be trained for a few epochs on a very small dataset. It will not be very good.
198
+
199
+ ### Demo
200
+
201
+ ```bash
202
+
203
+ # download and prepare the co3d subset
204
+ mkdir -p data/co3d_subset
205
+ cd data/co3d_subset
206
+ git clone https://github.com/facebookresearch/co3d
207
+ cd co3d
208
+ python3 ./co3d/download_dataset.py --download_folder ../ --single_sequence_subset
209
+ rm ../*.zip
210
+ cd ../../..
211
+
212
+ python3 datasets_preprocess/preprocess_co3d.py --co3d_dir data/co3d_subset --output_dir data/co3d_subset_processed --single_sequence_subset
213
+
214
+ # download the pretrained croco v2 checkpoint
215
+ mkdir -p checkpoints/
216
+ wget https://download.europe.naverlabs.com/ComputerVision/CroCo/CroCo_V2_ViTLarge_BaseDecoder.pth -P checkpoints/
217
+
218
+ # the training of dust3r is done in 3 steps.
219
+ # for this example we'll do fewer epochs, for the actual hyperparameters we used in the paper, see the next section: "Our Hyperparameters"
220
+ # step 1 - train dust3r for 224 resolution
221
+ torchrun --nproc_per_node=4 train.py \
222
+ --train_dataset "1000 @ Co3d(split='train', ROOT='data/co3d_subset_processed', aug_crop=16, mask_bg='rand', resolution=224, transform=ColorJitter)" \
223
+ --test_dataset "100 @ Co3d(split='test', ROOT='data/co3d_subset_processed', resolution=224, seed=777)" \
224
+ --model "AsymmetricCroCo3DStereo(pos_embed='RoPE100', img_size=(224, 224), head_type='linear', output_mode='pts3d', depth_mode=('exp', -inf, inf), conf_mode=('exp', 1, inf), enc_embed_dim=1024, enc_depth=24, enc_num_heads=16, dec_embed_dim=768, dec_depth=12, dec_num_heads=12)" \
225
+ --train_criterion "ConfLoss(Regr3D(L21, norm_mode='avg_dis'), alpha=0.2)" \
226
+ --test_criterion "Regr3D_ScaleShiftInv(L21, gt_scale=True)" \
227
+ --pretrained checkpoints/CroCo_V2_ViTLarge_BaseDecoder.pth \
228
+ --lr 0.0001 --min_lr 1e-06 --warmup_epochs 1 --epochs 10 --batch_size 16 --accum_iter 1 \
229
+ --save_freq 1 --keep_freq 5 --eval_freq 1 \
230
+ --output_dir checkpoints/dust3r_demo_224
231
+
232
+ # step 2 - train dust3r for 512 resolution
233
+ torchrun --nproc_per_node=4 train.py \
234
+ --train_dataset "1000 @ Co3d(split='train', ROOT='data/co3d_subset_processed', aug_crop=16, mask_bg='rand', resolution=[(512, 384), (512, 336), (512, 288), (512, 256), (512, 160)], transform=ColorJitter)" \
235
+ --test_dataset="100 @ Co3d(split='test', ROOT='data/co3d_subset_processed', resolution=(512,384), seed=777)" \
236
+ --model="AsymmetricCroCo3DStereo(pos_embed='RoPE100', patch_embed_cls='ManyAR_PatchEmbed', img_size=(512, 512), head_type='linear', output_mode='pts3d', depth_mode=('exp', -inf, inf), conf_mode=('exp', 1, inf), enc_embed_dim=1024, enc_depth=24, enc_num_heads=16, dec_embed_dim=768, dec_depth=12, dec_num_heads=12)" \
237
+ --train_criterion "ConfLoss(Regr3D(L21, norm_mode='avg_dis'), alpha=0.2)" \
238
+ --test_criterion "Regr3D_ScaleShiftInv(L21, gt_scale=True)" \
239
+ --pretrained='checkpoints/dust3r_demo_224/checkpoint-best.pth' \
240
+ --lr=0.0001 --min_lr=1e-06 --warmup_epochs 1 --epochs 10 --batch_size 4 --accum_iter 4 \
241
+ --save_freq 1 --keep_freq 5 --eval_freq 1 \
242
+ --output_dir checkpoints/dust3r_demo_512
243
+
244
+ # step 3 - train dust3r for 512 resolution with dpt
245
+ torchrun --nproc_per_node=4 train.py \
246
+ --train_dataset "1000 @ Co3d(split='train', ROOT='data/co3d_subset_processed', aug_crop=16, mask_bg='rand', resolution=[(512, 384), (512, 336), (512, 288), (512, 256), (512, 160)], transform=ColorJitter)" \
247
+ --test_dataset="100 @ Co3d(split='test', ROOT='data/co3d_subset_processed', resolution=(512,384), seed=777)" \
248
+ --model="AsymmetricCroCo3DStereo(pos_embed='RoPE100', patch_embed_cls='ManyAR_PatchEmbed', img_size=(512, 512), head_type='dpt', output_mode='pts3d', depth_mode=('exp', -inf, inf), conf_mode=('exp', 1, inf), enc_embed_dim=1024, enc_depth=24, enc_num_heads=16, dec_embed_dim=768, dec_depth=12, dec_num_heads=12)" \
249
+ --train_criterion "ConfLoss(Regr3D(L21, norm_mode='avg_dis'), alpha=0.2)" \
250
+ --test_criterion "Regr3D_ScaleShiftInv(L21, gt_scale=True)" \
251
+ --pretrained='checkpoints/dust3r_demo_512/checkpoint-best.pth' \
252
+ --lr=0.0001 --min_lr=1e-06 --warmup_epochs 1 --epochs 10 --batch_size 2 --accum_iter 8 \
253
+ --save_freq 1 --keep_freq 5 --eval_freq 1 \
254
+ --output_dir checkpoints/dust3r_demo_512dpt
255
+
256
+ ```
257
+
258
+ ### Our Hyperparameters
259
+ We didn't release the training datasets, but here are the commands we used for training our models:
260
+
261
+ ```bash
262
+ # NOTE: ROOT path omitted for datasets
263
+ # 224 linear
264
+ torchrun --nproc_per_node 4 train.py \
265
+ --train_dataset=" + 100_000 @ Habitat512(1_000_000, split='train', aug_crop=16, resolution=224, transform=ColorJitter) + 100_000 @ BlendedMVS(split='train', aug_crop=16, resolution=224, transform=ColorJitter) + 100_000 @ MegaDepthDense(split='train', aug_crop=16, resolution=224, transform=ColorJitter) + 100_000 @ ARKitScenes(aug_crop=256, resolution=224, transform=ColorJitter) + 100_000 @ Co3d_v3(split='train', aug_crop=16, mask_bg='rand', resolution=224, transform=ColorJitter) + 100_000 @ StaticThings3D(aug_crop=256, mask_bg='rand', resolution=224, transform=ColorJitter) + 100_000 @ ScanNetpp(split='train', aug_crop=256, resolution=224, transform=ColorJitter) + 100_000 @ Waymo(aug_crop=128, resolution=224, transform=ColorJitter) " \
266
+ --test_dataset=" Habitat512(1_000, split='val', resolution=224, seed=777) + 1_000 @ BlendedMVS(split='val', resolution=224, seed=777) + 1_000 @ MegaDepthDense(split='val', resolution=224, seed=777) + 1_000 @ Co3d_v3(split='test', mask_bg='rand', resolution=224, seed=777) " \
267
+ --train_criterion="ConfLoss(Regr3D(L21, norm_mode='avg_dis'), alpha=0.2)" \
268
+ --test_criterion='Regr3D_ScaleShiftInv(L21, gt_scale=True)' \
269
+ --model="AsymmetricCroCo3DStereo(pos_embed='RoPE100', img_size=(224, 224), head_type='linear', output_mode='pts3d', depth_mode=('exp', -inf, inf), conf_mode=('exp', 1, inf), enc_embed_dim=1024, enc_depth=24, enc_num_heads=16, dec_embed_dim=768, dec_depth=12, dec_num_heads=12)" \
270
+ --pretrained="checkpoints/CroCo_V2_ViTLarge_BaseDecoder.pth" \
271
+ --lr=0.0001 --min_lr=1e-06 --warmup_epochs=10 --epochs=100 --batch_size=16 --accum_iter=1 \
272
+ --save_freq=5 --keep_freq=10 --eval_freq=1 \
273
+ --output_dir='checkpoints/dust3r_224'
274
+
275
+ # 512 linear
276
+ torchrun --nproc_per_node 8 train.py \
277
+ --train_dataset=" + 10_000 @ Habitat512(1_000_000, split='train', aug_crop=16, resolution=[(512, 384), (512, 336), (512, 288), (512, 256), (512, 160)], transform=ColorJitter) + 10_000 @ BlendedMVS(split='train', aug_crop=16, resolution=[(512, 384), (512, 336), (512, 288), (512, 256), (512, 160)], transform=ColorJitter) + 10_000 @ MegaDepthDense(split='train', aug_crop=16, resolution=[(512, 384), (512, 336), (512, 288), (512, 256), (512, 160)], transform=ColorJitter) + 10_000 @ ARKitScenes(aug_crop=256, resolution=[(512, 384), (512, 336), (512, 288), (512, 256), (512, 160)], transform=ColorJitter) + 10_000 @ Co3d_v3(split='train', aug_crop=16, mask_bg='rand', resolution=[(512, 384), (512, 336), (512, 288), (512, 256), (512, 160)], transform=ColorJitter) + 10_000 @ StaticThings3D(aug_crop=256, mask_bg='rand', resolution=[(512, 384), (512, 336), (512, 288), (512, 256), (512, 160)], transform=ColorJitter) + 10_000 @ ScanNetpp(split='train', aug_crop=256, resolution=[(512, 384), (512, 336), (512, 288), (512, 256), (512, 160)], transform=ColorJitter) + 10_000 @ Waymo(aug_crop=128, resolution=[(512, 384), (512, 336), (512, 288), (512, 256), (512, 160)], transform=ColorJitter) " \
278
+ --test_dataset=" Habitat512(1_000, split='val', resolution=(512,384), seed=777) + 1_000 @ BlendedMVS(split='val', resolution=(512,384), seed=777) + 1_000 @ MegaDepthDense(split='val', resolution=(512,336), seed=777) + 1_000 @ Co3d_v3(split='test', resolution=(512,384), seed=777) " \
279
+ --train_criterion="ConfLoss(Regr3D(L21, norm_mode='avg_dis'), alpha=0.2)" \
280
+ --test_criterion='Regr3D_ScaleShiftInv(L21, gt_scale=True)' \
281
+ --model="AsymmetricCroCo3DStereo(pos_embed='RoPE100', patch_embed_cls='ManyAR_PatchEmbed', img_size=(512, 512), head_type='linear', output_mode='pts3d', depth_mode=('exp', -inf, inf), conf_mode=('exp', 1, inf), enc_embed_dim=1024, enc_depth=24, enc_num_heads=16, dec_embed_dim=768, dec_depth=12, dec_num_heads=12)" \
282
+ --pretrained='checkpoints/dust3r_224/checkpoint-best.pth' \
283
+ --lr=0.0001 --min_lr=1e-06 --warmup_epochs=20 --epochs=200 --batch_size=4 --accum_iter=2 \
284
+ --save_freq=10 --keep_freq=10 --eval_freq=1 --print_freq=10 \
285
+ --output_dir='checkpoints/dust3r_512'
286
+
287
+ # 512 dpt
288
+ torchrun --nproc_per_node 8 train.py \
289
+ --train_dataset=" + 10_000 @ Habitat512(1_000_000, split='train', aug_crop=16, resolution=[(512, 384), (512, 336), (512, 288), (512, 256), (512, 160)], transform=ColorJitter) + 10_000 @ BlendedMVS(split='train', aug_crop=16, resolution=[(512, 384), (512, 336), (512, 288), (512, 256), (512, 160)], transform=ColorJitter) + 10_000 @ MegaDepthDense(split='train', aug_crop=16, resolution=[(512, 384), (512, 336), (512, 288), (512, 256), (512, 160)], transform=ColorJitter) + 10_000 @ ARKitScenes(aug_crop=256, resolution=[(512, 384), (512, 336), (512, 288), (512, 256), (512, 160)], transform=ColorJitter) + 10_000 @ Co3d_v3(split='train', aug_crop=16, mask_bg='rand', resolution=[(512, 384), (512, 336), (512, 288), (512, 256), (512, 160)], transform=ColorJitter) + 10_000 @ StaticThings3D(aug_crop=256, mask_bg='rand', resolution=[(512, 384), (512, 336), (512, 288), (512, 256), (512, 160)], transform=ColorJitter) + 10_000 @ ScanNetpp(split='train', aug_crop=256, resolution=[(512, 384), (512, 336), (512, 288), (512, 256), (512, 160)], transform=ColorJitter) + 10_000 @ Waymo(aug_crop=128, resolution=[(512, 384), (512, 336), (512, 288), (512, 256), (512, 160)], transform=ColorJitter) " \
290
+ --test_dataset=" Habitat512(1_000, split='val', resolution=(512,384), seed=777) + 1_000 @ BlendedMVS(split='val', resolution=(512,384), seed=777) + 1_000 @ MegaDepthDense(split='val', resolution=(512,336), seed=777) + 1_000 @ Co3d_v3(split='test', resolution=(512,384), seed=777) " \
291
+ --train_criterion="ConfLoss(Regr3D(L21, norm_mode='avg_dis'), alpha=0.2)" \
292
+ --test_criterion='Regr3D_ScaleShiftInv(L21, gt_scale=True)' \
293
+ --model="AsymmetricCroCo3DStereo(pos_embed='RoPE100', patch_embed_cls='ManyAR_PatchEmbed', img_size=(512, 512), head_type='dpt', output_mode='pts3d', depth_mode=('exp', -inf, inf), conf_mode=('exp', 1, inf), enc_embed_dim=1024, enc_depth=24, enc_num_heads=16, dec_embed_dim=768, dec_depth=12, dec_num_heads=12)" \
294
+ --pretrained='checkpoints/dust3r_512/checkpoint-best.pth' \
295
+ --lr=0.0001 --min_lr=1e-06 --warmup_epochs=15 --epochs=90 --batch_size=2 --accum_iter=4 \
296
+ --save_freq=5 --keep_freq=10 --eval_freq=1 --print_freq=10 \
297
+ --output_dir='checkpoints/dust3r_512dpt'
298
+
299
+ ```
demo/dust3r/assets/demo.jpg ADDED
demo/dust3r/assets/dust3r_archi.jpg ADDED
demo/dust3r/assets/matching.jpg ADDED
demo/dust3r/assets/pipeline1.jpg ADDED
demo/dust3r/checkpoints/DUSt3R_ViTLarge_BaseDecoder_512_dpt.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:15934abef92c7071953f65cb858b23b52737f1dee7934198f4abca67f6eb8949
3
+ size 407543808
demo/dust3r/checkpoints/DUSt3R_ViTLarge_BaseDecoder_512_dpt.pth.1 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5e8bbf0c4d1d6007f5343f3f45814b956ddc5bbb4d00cb66beaf73afe5c53b34
3
+ size 2285019929
demo/dust3r/croco/LICENSE ADDED
@@ -0,0 +1,52 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ CroCo, Copyright (c) 2022-present Naver Corporation, is licensed under the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 license.
2
+
3
+ A summary of the CC BY-NC-SA 4.0 license is located here:
4
+ https://creativecommons.org/licenses/by-nc-sa/4.0/
5
+
6
+ The CC BY-NC-SA 4.0 license is located here:
7
+ https://creativecommons.org/licenses/by-nc-sa/4.0/legalcode
8
+
9
+
10
+ SEE NOTICE BELOW WITH RESPECT TO THE FILE: models/pos_embed.py, models/blocks.py
11
+
12
+ ***************************
13
+
14
+ NOTICE WITH RESPECT TO THE FILE: models/pos_embed.py
15
+
16
+ This software is being redistributed in a modifiled form. The original form is available here:
17
+
18
+ https://github.com/facebookresearch/mae/blob/main/util/pos_embed.py
19
+
20
+ This software in this file incorporates parts of the following software available here:
21
+
22
+ Transformer: https://github.com/tensorflow/models/blob/master/official/legacy/transformer/model_utils.py
23
+ available under the following license: https://github.com/tensorflow/models/blob/master/LICENSE
24
+
25
+ MoCo v3: https://github.com/facebookresearch/moco-v3
26
+ available under the following license: https://github.com/facebookresearch/moco-v3/blob/main/LICENSE
27
+
28
+ DeiT: https://github.com/facebookresearch/deit
29
+ available under the following license: https://github.com/facebookresearch/deit/blob/main/LICENSE
30
+
31
+
32
+ ORIGINAL COPYRIGHT NOTICE AND PERMISSION NOTICE AVAILABLE HERE IS REPRODUCE BELOW:
33
+
34
+ https://github.com/facebookresearch/mae/blob/main/LICENSE
35
+
36
+ Attribution-NonCommercial 4.0 International
37
+
38
+ ***************************
39
+
40
+ NOTICE WITH RESPECT TO THE FILE: models/blocks.py
41
+
42
+ This software is being redistributed in a modifiled form. The original form is available here:
43
+
44
+ https://github.com/rwightman/pytorch-image-models
45
+
46
+ ORIGINAL COPYRIGHT NOTICE AND PERMISSION NOTICE AVAILABLE HERE IS REPRODUCE BELOW:
47
+
48
+ https://github.com/rwightman/pytorch-image-models/blob/master/LICENSE
49
+
50
+ Apache License
51
+ Version 2.0, January 2004
52
+ http://www.apache.org/licenses/
demo/dust3r/croco/NOTICE ADDED
@@ -0,0 +1,21 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ CroCo
2
+ Copyright 2022-present NAVER Corp.
3
+
4
+ This project contains subcomponents with separate copyright notices and license terms.
5
+ Your use of the source code for these subcomponents is subject to the terms and conditions of the following licenses.
6
+
7
+ ====
8
+
9
+ facebookresearch/mae
10
+ https://github.com/facebookresearch/mae
11
+
12
+ Attribution-NonCommercial 4.0 International
13
+
14
+ ====
15
+
16
+ rwightman/pytorch-image-models
17
+ https://github.com/rwightman/pytorch-image-models
18
+
19
+ Apache License
20
+ Version 2.0, January 2004
21
+ http://www.apache.org/licenses/
demo/dust3r/croco/README.MD ADDED
@@ -0,0 +1,124 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # CroCo + CroCo v2 / CroCo-Stereo / CroCo-Flow
2
+
3
+ [[`CroCo arXiv`](https://arxiv.org/abs/2210.10716)] [[`CroCo v2 arXiv`](https://arxiv.org/abs/2211.10408)] [[`project page and demo`](https://croco.europe.naverlabs.com/)]
4
+
5
+ This repository contains the code for our CroCo model presented in our NeurIPS'22 paper [CroCo: Self-Supervised Pre-training for 3D Vision Tasks by Cross-View Completion](https://openreview.net/pdf?id=wZEfHUM5ri) and its follow-up extension published at ICCV'23 [Improved Cross-view Completion Pre-training for Stereo Matching and Optical Flow](https://openaccess.thecvf.com/content/ICCV2023/html/Weinzaepfel_CroCo_v2_Improved_Cross-view_Completion_Pre-training_for_Stereo_Matching_and_ICCV_2023_paper.html), refered to as CroCo v2:
6
+
7
+ ![image](assets/arch.jpg)
8
+
9
+ ```bibtex
10
+ @inproceedings{croco,
11
+ title={{CroCo: Self-Supervised Pre-training for 3D Vision Tasks by Cross-View Completion}},
12
+ author={{Weinzaepfel, Philippe and Leroy, Vincent and Lucas, Thomas and Br\'egier, Romain and Cabon, Yohann and Arora, Vaibhav and Antsfeld, Leonid and Chidlovskii, Boris and Csurka, Gabriela and Revaud J\'er\^ome}},
13
+ booktitle={{NeurIPS}},
14
+ year={2022}
15
+ }
16
+
17
+ @inproceedings{croco_v2,
18
+ title={{CroCo v2: Improved Cross-view Completion Pre-training for Stereo Matching and Optical Flow}},
19
+ author={Weinzaepfel, Philippe and Lucas, Thomas and Leroy, Vincent and Cabon, Yohann and Arora, Vaibhav and Br{\'e}gier, Romain and Csurka, Gabriela and Antsfeld, Leonid and Chidlovskii, Boris and Revaud, J{\'e}r{\^o}me},
20
+ booktitle={ICCV},
21
+ year={2023}
22
+ }
23
+ ```
24
+
25
+ ## License
26
+
27
+ The code is distributed under the CC BY-NC-SA 4.0 License. See [LICENSE](LICENSE) for more information.
28
+ Some components are based on code from [MAE](https://github.com/facebookresearch/mae) released under the CC BY-NC-SA 4.0 License and [timm](https://github.com/rwightman/pytorch-image-models) released under the Apache 2.0 License.
29
+ Some components for stereo matching and optical flow are based on code from [unimatch](https://github.com/autonomousvision/unimatch) released under the MIT license.
30
+
31
+ ## Preparation
32
+
33
+ 1. Install dependencies on a machine with a NVidia GPU using e.g. conda. Note that `habitat-sim` is required only for the interactive demo and the synthetic pre-training data generation. If you don't plan to use it, you can ignore the line installing it and use a more recent python version.
34
+
35
+ ```bash
36
+ conda create -n croco python=3.7 cmake=3.14.0
37
+ conda activate croco
38
+ conda install habitat-sim headless -c conda-forge -c aihabitat
39
+ conda install pytorch torchvision -c pytorch
40
+ conda install notebook ipykernel matplotlib
41
+ conda install ipywidgets widgetsnbextension
42
+ conda install scikit-learn tqdm quaternion opencv # only for pretraining / habitat data generation
43
+
44
+ ```
45
+
46
+ 2. Compile cuda kernels for RoPE
47
+
48
+ CroCo v2 relies on RoPE positional embeddings for which you need to compile some cuda kernels.
49
+ ```bash
50
+ cd models/curope/
51
+ python setup.py build_ext --inplace
52
+ cd ../../
53
+ ```
54
+
55
+ This can be a bit long as we compile for all cuda architectures, feel free to update L9 of `models/curope/setup.py` to compile for specific architectures only.
56
+ You might also need to set the environment `CUDA_HOME` in case you use a custom cuda installation.
57
+
58
+ In case you cannot provide, we also provide a slow pytorch version, which will be automatically loaded.
59
+
60
+ 3. Download pre-trained model
61
+
62
+ We provide several pre-trained models:
63
+
64
+ | modelname | pre-training data | pos. embed. | Encoder | Decoder |
65
+ |------------------------------------------------------------------------------------------------------------------------------------|-------------------|-------------|---------|---------|
66
+ | [`CroCo.pth`](https://download.europe.naverlabs.com/ComputerVision/CroCo/CroCo.pth) | Habitat | cosine | ViT-B | Small |
67
+ | [`CroCo_V2_ViTBase_SmallDecoder.pth`](https://download.europe.naverlabs.com/ComputerVision/CroCo/CroCo_V2_ViTBase_SmallDecoder.pth) | Habitat + real | RoPE | ViT-B | Small |
68
+ | [`CroCo_V2_ViTBase_BaseDecoder.pth`](https://download.europe.naverlabs.com/ComputerVision/CroCo/CroCo_V2_ViTBase_BaseDecoder.pth) | Habitat + real | RoPE | ViT-B | Base |
69
+ | [`CroCo_V2_ViTLarge_BaseDecoder.pth`](https://download.europe.naverlabs.com/ComputerVision/CroCo/CroCo_V2_ViTLarge_BaseDecoder.pth) | Habitat + real | RoPE | ViT-L | Base |
70
+
71
+ To download a specific model, i.e., the first one (`CroCo.pth`)
72
+ ```bash
73
+ mkdir -p pretrained_models/
74
+ wget https://download.europe.naverlabs.com/ComputerVision/CroCo/CroCo.pth -P pretrained_models/
75
+ ```
76
+
77
+ ## Reconstruction example
78
+
79
+ Simply run after downloading the `CroCo_V2_ViTLarge_BaseDecoder` pretrained model (or update the corresponding line in `demo.py`)
80
+ ```bash
81
+ python demo.py
82
+ ```
83
+
84
+ ## Interactive demonstration of cross-view completion reconstruction on the Habitat simulator
85
+
86
+ First download the test scene from Habitat:
87
+ ```bash
88
+ python -m habitat_sim.utils.datasets_download --uids habitat_test_scenes --data-path habitat-sim-data/
89
+ ```
90
+
91
+ Then, run the Notebook demo `interactive_demo.ipynb`.
92
+
93
+ In this demo, you should be able to sample a random reference viewpoint from an [Habitat](https://github.com/facebookresearch/habitat-sim) test scene. Use the sliders to change viewpoint and select a masked target view to reconstruct using CroCo.
94
+ ![croco_interactive_demo](https://user-images.githubusercontent.com/1822210/200516576-7937bc6a-55f8-49ed-8618-3ddf89433ea4.jpg)
95
+
96
+ ## Pre-training
97
+
98
+ ### CroCo
99
+
100
+ To pre-train CroCo, please first generate the pre-training data from the Habitat simulator, following the instructions in [datasets/habitat_sim/README.MD](datasets/habitat_sim/README.MD) and then run the following command:
101
+ ```
102
+ torchrun --nproc_per_node=4 pretrain.py --output_dir ./output/pretraining/
103
+ ```
104
+
105
+ Our CroCo pre-training was launched on a single server with 4 GPUs.
106
+ It should take around 10 days with A100 or 15 days with V100 to do the 400 pre-training epochs, but decent performances are obtained earlier in training.
107
+ Note that, while the code contains the same scaling rule of the learning rate as MAE when changing the effective batch size, we did not experimented if it is valid in our case.
108
+ The first run can take a few minutes to start, to parse all available pre-training pairs.
109
+
110
+ ### CroCo v2
111
+
112
+ For CroCo v2 pre-training, in addition to the generation of the pre-training data from the Habitat simulator above, please pre-extract the crops from the real datasets following the instructions in [datasets/crops/README.MD](datasets/crops/README.MD).
113
+ Then, run the following command for the largest model (ViT-L encoder, Base decoder):
114
+ ```
115
+ torchrun --nproc_per_node=8 pretrain.py --model "CroCoNet(enc_embed_dim=1024, enc_depth=24, enc_num_heads=16, dec_embed_dim=768, dec_num_heads=12, dec_depth=12, pos_embed='RoPE100')" --dataset "habitat_release+ARKitScenes+MegaDepth+3DStreetView+IndoorVL" --warmup_epochs 12 --max_epoch 125 --epochs 250 --amp 0 --keep_freq 5 --output_dir ./output/pretraining_crocov2/
116
+ ```
117
+
118
+ Our CroCo v2 pre-training was launched on a single server with 8 GPUs for the largest model, and on a single server with 4 GPUs for the smaller ones, keeping a batch size of 64 per gpu in all cases.
119
+ The largest model should take around 12 days on A100.
120
+ Note that, while the code contains the same scaling rule of the learning rate as MAE when changing the effective batch size, we did not experimented if it is valid in our case.
121
+
122
+ ## Stereo matching and Optical flow downstream tasks
123
+
124
+ For CroCo-Stereo and CroCo-Flow, please refer to [stereoflow/README.MD](stereoflow/README.MD).
demo/dust3r/croco/assets/Chateau1.png ADDED
demo/dust3r/croco/assets/Chateau2.png ADDED
demo/dust3r/croco/assets/arch.jpg ADDED
demo/dust3r/croco/croco-stereo-flow-demo.ipynb ADDED
@@ -0,0 +1,191 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "cells": [
3
+ {
4
+ "cell_type": "markdown",
5
+ "id": "9bca0f41",
6
+ "metadata": {},
7
+ "source": [
8
+ "# Simple inference example with CroCo-Stereo or CroCo-Flow"
9
+ ]
10
+ },
11
+ {
12
+ "cell_type": "code",
13
+ "execution_count": null,
14
+ "id": "80653ef7",
15
+ "metadata": {},
16
+ "outputs": [],
17
+ "source": [
18
+ "# Copyright (C) 2022-present Naver Corporation. All rights reserved.\n",
19
+ "# Licensed under CC BY-NC-SA 4.0 (non-commercial use only)."
20
+ ]
21
+ },
22
+ {
23
+ "cell_type": "markdown",
24
+ "id": "4f033862",
25
+ "metadata": {},
26
+ "source": [
27
+ "First download the model(s) of your choice by running\n",
28
+ "```\n",
29
+ "bash stereoflow/download_model.sh crocostereo.pth\n",
30
+ "bash stereoflow/download_model.sh crocoflow.pth\n",
31
+ "```"
32
+ ]
33
+ },
34
+ {
35
+ "cell_type": "code",
36
+ "execution_count": null,
37
+ "id": "1fb2e392",
38
+ "metadata": {},
39
+ "outputs": [],
40
+ "source": [
41
+ "import torch\n",
42
+ "use_gpu = torch.cuda.is_available() and torch.cuda.device_count()>0\n",
43
+ "device = torch.device('cuda:0' if use_gpu else 'cpu')\n",
44
+ "import matplotlib.pylab as plt"
45
+ ]
46
+ },
47
+ {
48
+ "cell_type": "code",
49
+ "execution_count": null,
50
+ "id": "e0e25d77",
51
+ "metadata": {},
52
+ "outputs": [],
53
+ "source": [
54
+ "from stereoflow.test import _load_model_and_criterion\n",
55
+ "from stereoflow.engine import tiled_pred\n",
56
+ "from stereoflow.datasets_stereo import img_to_tensor, vis_disparity\n",
57
+ "from stereoflow.datasets_flow import flowToColor\n",
58
+ "tile_overlap=0.7 # recommended value, higher value can be slightly better but slower"
59
+ ]
60
+ },
61
+ {
62
+ "cell_type": "markdown",
63
+ "id": "86a921f5",
64
+ "metadata": {},
65
+ "source": [
66
+ "### CroCo-Stereo example"
67
+ ]
68
+ },
69
+ {
70
+ "cell_type": "code",
71
+ "execution_count": null,
72
+ "id": "64e483cb",
73
+ "metadata": {},
74
+ "outputs": [],
75
+ "source": [
76
+ "image1 = np.asarray(Image.open('<path_to_left_image>'))\n",
77
+ "image2 = np.asarray(Image.open('<path_to_right_image>'))"
78
+ ]
79
+ },
80
+ {
81
+ "cell_type": "code",
82
+ "execution_count": null,
83
+ "id": "f0d04303",
84
+ "metadata": {},
85
+ "outputs": [],
86
+ "source": [
87
+ "model, _, cropsize, with_conf, task, tile_conf_mode = _load_model_and_criterion('stereoflow_models/crocostereo.pth', None, device)\n"
88
+ ]
89
+ },
90
+ {
91
+ "cell_type": "code",
92
+ "execution_count": null,
93
+ "id": "47dc14b5",
94
+ "metadata": {},
95
+ "outputs": [],
96
+ "source": [
97
+ "im1 = img_to_tensor(image1).to(device).unsqueeze(0)\n",
98
+ "im2 = img_to_tensor(image2).to(device).unsqueeze(0)\n",
99
+ "with torch.inference_mode():\n",
100
+ " pred, _, _ = tiled_pred(model, None, im1, im2, None, conf_mode=tile_conf_mode, overlap=tile_overlap, crop=cropsize, with_conf=with_conf, return_time=False)\n",
101
+ "pred = pred.squeeze(0).squeeze(0).cpu().numpy()"
102
+ ]
103
+ },
104
+ {
105
+ "cell_type": "code",
106
+ "execution_count": null,
107
+ "id": "583b9f16",
108
+ "metadata": {},
109
+ "outputs": [],
110
+ "source": [
111
+ "plt.imshow(vis_disparity(pred))\n",
112
+ "plt.axis('off')"
113
+ ]
114
+ },
115
+ {
116
+ "cell_type": "markdown",
117
+ "id": "d2df5d70",
118
+ "metadata": {},
119
+ "source": [
120
+ "### CroCo-Flow example"
121
+ ]
122
+ },
123
+ {
124
+ "cell_type": "code",
125
+ "execution_count": null,
126
+ "id": "9ee257a7",
127
+ "metadata": {},
128
+ "outputs": [],
129
+ "source": [
130
+ "image1 = np.asarray(Image.open('<path_to_first_image>'))\n",
131
+ "image2 = np.asarray(Image.open('<path_to_second_image>'))"
132
+ ]
133
+ },
134
+ {
135
+ "cell_type": "code",
136
+ "execution_count": null,
137
+ "id": "d5edccf0",
138
+ "metadata": {},
139
+ "outputs": [],
140
+ "source": [
141
+ "model, _, cropsize, with_conf, task, tile_conf_mode = _load_model_and_criterion('stereoflow_models/crocoflow.pth', None, device)\n"
142
+ ]
143
+ },
144
+ {
145
+ "cell_type": "code",
146
+ "execution_count": null,
147
+ "id": "b19692c3",
148
+ "metadata": {},
149
+ "outputs": [],
150
+ "source": [
151
+ "im1 = img_to_tensor(image1).to(device).unsqueeze(0)\n",
152
+ "im2 = img_to_tensor(image2).to(device).unsqueeze(0)\n",
153
+ "with torch.inference_mode():\n",
154
+ " pred, _, _ = tiled_pred(model, None, im1, im2, None, conf_mode=tile_conf_mode, overlap=tile_overlap, crop=cropsize, with_conf=with_conf, return_time=False)\n",
155
+ "pred = pred.squeeze(0).permute(1,2,0).cpu().numpy()"
156
+ ]
157
+ },
158
+ {
159
+ "cell_type": "code",
160
+ "execution_count": null,
161
+ "id": "26f79db3",
162
+ "metadata": {},
163
+ "outputs": [],
164
+ "source": [
165
+ "plt.imshow(flowToColor(pred))\n",
166
+ "plt.axis('off')"
167
+ ]
168
+ }
169
+ ],
170
+ "metadata": {
171
+ "kernelspec": {
172
+ "display_name": "Python 3 (ipykernel)",
173
+ "language": "python",
174
+ "name": "python3"
175
+ },
176
+ "language_info": {
177
+ "codemirror_mode": {
178
+ "name": "ipython",
179
+ "version": 3
180
+ },
181
+ "file_extension": ".py",
182
+ "mimetype": "text/x-python",
183
+ "name": "python",
184
+ "nbconvert_exporter": "python",
185
+ "pygments_lexer": "ipython3",
186
+ "version": "3.9.7"
187
+ }
188
+ },
189
+ "nbformat": 4,
190
+ "nbformat_minor": 5
191
+ }
demo/dust3r/croco/datasets/__init__.py ADDED
File without changes
demo/dust3r/croco/datasets/crops/README.MD ADDED
@@ -0,0 +1,104 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ## Generation of crops from the real datasets
2
+
3
+ The instructions below allow to generate the crops used for pre-training CroCo v2 from the following real-world datasets: ARKitScenes, MegaDepth, 3DStreetView and IndoorVL.
4
+
5
+ ### Download the metadata of the crops to generate
6
+
7
+ First, download the metadata and put them in `./data/`:
8
+ ```
9
+ mkdir -p data
10
+ cd data/
11
+ wget https://download.europe.naverlabs.com/ComputerVision/CroCo/data/crop_metadata.zip
12
+ unzip crop_metadata.zip
13
+ rm crop_metadata.zip
14
+ cd ..
15
+ ```
16
+
17
+ ### Prepare the original datasets
18
+
19
+ Second, download the original datasets in `./data/original_datasets/`.
20
+ ```
21
+ mkdir -p data/original_datasets
22
+ ```
23
+
24
+ ##### ARKitScenes
25
+
26
+ Download the `raw` dataset from https://github.com/apple/ARKitScenes/blob/main/DATA.md and put it in `./data/original_datasets/ARKitScenes/`.
27
+ The resulting file structure should be like:
28
+ ```
29
+ ./data/original_datasets/ARKitScenes/
30
+ └───Training
31
+ └───40753679
32
+ │ │ ultrawide
33
+ │ │ ...
34
+ └───40753686
35
+ │
36
+ ...
37
+ ```
38
+
39
+ ##### MegaDepth
40
+
41
+ Download `MegaDepth v1 Dataset` from https://www.cs.cornell.edu/projects/megadepth/ and put it in `./data/original_datasets/MegaDepth/`.
42
+ The resulting file structure should be like:
43
+
44
+ ```
45
+ ./data/original_datasets/MegaDepth/
46
+ └───0000
47
+ │ └───images
48
+ │ │ │ 1000557903_87fa96b8a4_o.jpg
49
+ │ │ └ ...
50
+ │ └─── ...
51
+ └───0001
52
+ │ │
53
+ │ └ ...
54
+ └─── ...
55
+ ```
56
+
57
+ ##### 3DStreetView
58
+
59
+ Download `3D_Street_View` dataset from https://github.com/amir32002/3D_Street_View and put it in `./data/original_datasets/3DStreetView/`.
60
+ The resulting file structure should be like:
61
+
62
+ ```
63
+ ./data/original_datasets/3DStreetView/
64
+ └───dataset_aligned
65
+ │ └───0002
66
+ │ │ │ 0000002_0000001_0000002_0000001.jpg
67
+ │ │ └ ...
68
+ │ └─── ...
69
+ └───dataset_unaligned
70
+ │ └───0003
71
+ │ │ │ 0000003_0000001_0000002_0000001.jpg
72
+ │ │ └ ...
73
+ │ └─── ...
74
+ ```
75
+
76
+ ##### IndoorVL
77
+
78
+ Download the `IndoorVL` datasets using [Kapture](https://github.com/naver/kapture).
79
+
80
+ ```
81
+ pip install kapture
82
+ mkdir -p ./data/original_datasets/IndoorVL
83
+ cd ./data/original_datasets/IndoorVL
84
+ kapture_download_dataset.py update
85
+ kapture_download_dataset.py install "HyundaiDepartmentStore_*"
86
+ kapture_download_dataset.py install "GangnamStation_*"
87
+ cd -
88
+ ```
89
+
90
+ ### Extract the crops
91
+
92
+ Now, extract the crops for each of the dataset:
93
+ ```
94
+ for dataset in ARKitScenes MegaDepth 3DStreetView IndoorVL;
95
+ do
96
+ python3 datasets/crops/extract_crops_from_images.py --crops ./data/crop_metadata/${dataset}/crops_release.txt --root-dir ./data/original_datasets/${dataset}/ --output-dir ./data/${dataset}_crops/ --imsize 256 --nthread 8 --max-subdir-levels 5 --ideal-number-pairs-in-dir 500;
97
+ done
98
+ ```
99
+
100
+ ##### Note for IndoorVL
101
+
102
+ Due to some legal issues, we can only release 144,228 pairs out of the 1,593,689 pairs used in the paper.
103
+ To account for it in terms of number of pre-training iterations, the pre-training command in this repository uses 125 training epochs including 12 warm-up epochs and learning rate cosine schedule of 250, instead of 100, 10 and 200 respectively.
104
+ The impact on the performance is negligible.
demo/dust3r/croco/datasets/crops/extract_crops_from_images.py ADDED
@@ -0,0 +1,159 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Copyright (C) 2022-present Naver Corporation. All rights reserved.
2
+ # Licensed under CC BY-NC-SA 4.0 (non-commercial use only).
3
+ #
4
+ # --------------------------------------------------------
5
+ # Extracting crops for pre-training
6
+ # --------------------------------------------------------
7
+
8
+ import os
9
+ import argparse
10
+ from tqdm import tqdm
11
+ from PIL import Image
12
+ import functools
13
+ from multiprocessing import Pool
14
+ import math
15
+
16
+
17
+ def arg_parser():
18
+ parser = argparse.ArgumentParser('Generate cropped image pairs from image crop list')
19
+
20
+ parser.add_argument('--crops', type=str, required=True, help='crop file')
21
+ parser.add_argument('--root-dir', type=str, required=True, help='root directory')
22
+ parser.add_argument('--output-dir', type=str, required=True, help='output directory')
23
+ parser.add_argument('--imsize', type=int, default=256, help='size of the crops')
24
+ parser.add_argument('--nthread', type=int, required=True, help='number of simultaneous threads')
25
+ parser.add_argument('--max-subdir-levels', type=int, default=5, help='maximum number of subdirectories')
26
+ parser.add_argument('--ideal-number-pairs-in-dir', type=int, default=500, help='number of pairs stored in a dir')
27
+ return parser
28
+
29
+
30
+ def main(args):
31
+ listing_path = os.path.join(args.output_dir, 'listing.txt')
32
+
33
+ print(f'Loading list of crops ... ({args.nthread} threads)')
34
+ crops, num_crops_to_generate = load_crop_file(args.crops)
35
+
36
+ print(f'Preparing jobs ({len(crops)} candidate image pairs)...')
37
+ num_levels = min(math.ceil(math.log(num_crops_to_generate, args.ideal_number_pairs_in_dir)), args.max_subdir_levels)
38
+ num_pairs_in_dir = math.ceil(num_crops_to_generate ** (1/num_levels))
39
+
40
+ jobs = prepare_jobs(crops, num_levels, num_pairs_in_dir)
41
+ del crops
42
+
43
+ os.makedirs(args.output_dir, exist_ok=True)
44
+ mmap = Pool(args.nthread).imap_unordered if args.nthread > 1 else map
45
+ call = functools.partial(save_image_crops, args)
46
+
47
+ print(f"Generating cropped images to {args.output_dir} ...")
48
+ with open(listing_path, 'w') as listing:
49
+ listing.write('# pair_path\n')
50
+ for results in tqdm(mmap(call, jobs), total=len(jobs)):
51
+ for path in results:
52
+ listing.write(f'{path}\n')
53
+ print('Finished writing listing to', listing_path)
54
+
55
+
56
+ def load_crop_file(path):
57
+ data = open(path).read().splitlines()
58
+ pairs = []
59
+ num_crops_to_generate = 0
60
+ for line in tqdm(data):
61
+ if line.startswith('#'):
62
+ continue
63
+ line = line.split(', ')
64
+ if len(line) < 8:
65
+ img1, img2, rotation = line
66
+ pairs.append((img1, img2, int(rotation), []))
67
+ else:
68
+ l1, r1, t1, b1, l2, r2, t2, b2 = map(int, line)
69
+ rect1, rect2 = (l1, t1, r1, b1), (l2, t2, r2, b2)
70
+ pairs[-1][-1].append((rect1, rect2))
71
+ num_crops_to_generate += 1
72
+ return pairs, num_crops_to_generate
73
+
74
+
75
+ def prepare_jobs(pairs, num_levels, num_pairs_in_dir):
76
+ jobs = []
77
+ powers = [num_pairs_in_dir**level for level in reversed(range(num_levels))]
78
+
79
+ def get_path(idx):
80
+ idx_array = []
81
+ d = idx
82
+ for level in range(num_levels - 1):
83
+ idx_array.append(idx // powers[level])
84
+ idx = idx % powers[level]
85
+ idx_array.append(d)
86
+ return '/'.join(map(lambda x: hex(x)[2:], idx_array))
87
+
88
+ idx = 0
89
+ for pair_data in tqdm(pairs):
90
+ img1, img2, rotation, crops = pair_data
91
+ if -60 <= rotation and rotation <= 60:
92
+ rotation = 0 # most likely not a true rotation
93
+ paths = [get_path(idx + k) for k in range(len(crops))]
94
+ idx += len(crops)
95
+ jobs.append(((img1, img2), rotation, crops, paths))
96
+ return jobs
97
+
98
+
99
+ def load_image(path):
100
+ try:
101
+ return Image.open(path).convert('RGB')
102
+ except Exception as e:
103
+ print('skipping', path, e)
104
+ raise OSError()
105
+
106
+
107
+ def save_image_crops(args, data):
108
+ # load images
109
+ img_pair, rot, crops, paths = data
110
+ try:
111
+ img1, img2 = [load_image(os.path.join(args.root_dir, impath)) for impath in img_pair]
112
+ except OSError as e:
113
+ return []
114
+
115
+ def area(sz):
116
+ return sz[0] * sz[1]
117
+
118
+ tgt_size = (args.imsize, args.imsize)
119
+
120
+ def prepare_crop(img, rect, rot=0):
121
+ # actual crop
122
+ img = img.crop(rect)
123
+
124
+ # resize to desired size
125
+ interp = Image.Resampling.LANCZOS if area(img.size) > 4*area(tgt_size) else Image.Resampling.BICUBIC
126
+ img = img.resize(tgt_size, resample=interp)
127
+
128
+ # rotate the image
129
+ rot90 = (round(rot/90) % 4) * 90
130
+ if rot90 == 90:
131
+ img = img.transpose(Image.Transpose.ROTATE_90)
132
+ elif rot90 == 180:
133
+ img = img.transpose(Image.Transpose.ROTATE_180)
134
+ elif rot90 == 270:
135
+ img = img.transpose(Image.Transpose.ROTATE_270)
136
+ return img
137
+
138
+ results = []
139
+ for (rect1, rect2), path in zip(crops, paths):
140
+ crop1 = prepare_crop(img1, rect1)
141
+ crop2 = prepare_crop(img2, rect2, rot)
142
+
143
+ fullpath1 = os.path.join(args.output_dir, path+'_1.jpg')
144
+ fullpath2 = os.path.join(args.output_dir, path+'_2.jpg')
145
+ os.makedirs(os.path.dirname(fullpath1), exist_ok=True)
146
+
147
+ assert not os.path.isfile(fullpath1), fullpath1
148
+ assert not os.path.isfile(fullpath2), fullpath2
149
+ crop1.save(fullpath1)
150
+ crop2.save(fullpath2)
151
+ results.append(path)
152
+
153
+ return results
154
+
155
+
156
+ if __name__ == '__main__':
157
+ args = arg_parser().parse_args()
158
+ main(args)
159
+
demo/dust3r/croco/datasets/habitat_sim/README.MD ADDED
@@ -0,0 +1,76 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ## Generation of synthetic image pairs using Habitat-Sim
2
+
3
+ These instructions allow to generate pre-training pairs from the Habitat simulator.
4
+ As we did not save metadata of the pairs used in the original paper, they are not strictly the same, but these data use the same setting and are equivalent.
5
+
6
+ ### Download Habitat-Sim scenes
7
+ Download Habitat-Sim scenes:
8
+ - Download links can be found here: https://github.com/facebookresearch/habitat-sim/blob/main/DATASETS.md
9
+ - We used scenes from the HM3D, habitat-test-scenes, Replica, ReplicaCad and ScanNet datasets.
10
+ - Please put the scenes under `./data/habitat-sim-data/scene_datasets/` following the structure below, or update manually paths in `paths.py`.
11
+ ```
12
+ ./data/
13
+ └──habitat-sim-data/
14
+ └──scene_datasets/
15
+ ├──hm3d/
16
+ ├──gibson/
17
+ ├──habitat-test-scenes/
18
+ ├──replica_cad_baked_lighting/
19
+ ├──replica_cad/
20
+ ├──ReplicaDataset/
21
+ └──scannet/
22
+ ```
23
+
24
+ ### Image pairs generation
25
+ We provide metadata to generate reproducible images pairs for pretraining and validation.
26
+ Experiments described in the paper used similar data, but whose generation was not reproducible at the time.
27
+
28
+ Specifications:
29
+ - 256x256 resolution images, with 60 degrees field of view .
30
+ - Up to 1000 image pairs per scene.
31
+ - Number of scenes considered/number of images pairs per dataset:
32
+ - Scannet: 1097 scenes / 985 209 pairs
33
+ - HM3D:
34
+ - hm3d/train: 800 / 800k pairs
35
+ - hm3d/val: 100 scenes / 100k pairs
36
+ - hm3d/minival: 10 scenes / 10k pairs
37
+ - habitat-test-scenes: 3 scenes / 3k pairs
38
+ - replica_cad_baked_lighting: 13 scenes / 13k pairs
39
+
40
+ - Scenes from hm3d/val and hm3d/minival pairs were not used for the pre-training but kept for validation purposes.
41
+
42
+ Download metadata and extract it:
43
+ ```bash
44
+ mkdir -p data/habitat_release_metadata/
45
+ cd data/habitat_release_metadata/
46
+ wget https://download.europe.naverlabs.com/ComputerVision/CroCo/data/habitat_release_metadata/multiview_habitat_metadata.tar.gz
47
+ tar -xvf multiview_habitat_metadata.tar.gz
48
+ cd ../..
49
+ # Location of the metadata
50
+ METADATA_DIR="./data/habitat_release_metadata/multiview_habitat_metadata"
51
+ ```
52
+
53
+ Generate image pairs from metadata:
54
+ - The following command will print a list of commandlines to generate image pairs for each scene:
55
+ ```bash
56
+ # Target output directory
57
+ PAIRS_DATASET_DIR="./data/habitat_release/"
58
+ python datasets/habitat_sim/generate_from_metadata_files.py --input_dir=$METADATA_DIR --output_dir=$PAIRS_DATASET_DIR
59
+ ```
60
+ - One can launch multiple of such commands in parallel e.g. using GNU Parallel:
61
+ ```bash
62
+ python datasets/habitat_sim/generate_from_metadata_files.py --input_dir=$METADATA_DIR --output_dir=$PAIRS_DATASET_DIR | parallel -j 16
63
+ ```
64
+
65
+ ## Metadata generation
66
+
67
+ Image pairs were randomly sampled using the following commands, whose outputs contain randomness and are thus not exactly reproducible:
68
+ ```bash
69
+ # Print commandlines to generate image pairs from the different scenes available.
70
+ PAIRS_DATASET_DIR=MY_CUSTOM_PATH
71
+ python datasets/habitat_sim/generate_multiview_images.py --list_commands --output_dir=$PAIRS_DATASET_DIR
72
+
73
+ # Once a dataset is generated, pack metadata files for reproducibility.
74
+ METADATA_DIR=MY_CUSTON_PATH
75
+ python datasets/habitat_sim/pack_metadata_files.py $PAIRS_DATASET_DIR $METADATA_DIR
76
+ ```
demo/dust3r/croco/datasets/habitat_sim/__init__.py ADDED
File without changes
demo/dust3r/croco/datasets/habitat_sim/generate_from_metadata.py ADDED
@@ -0,0 +1,92 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Copyright (C) 2022-present Naver Corporation. All rights reserved.
2
+ # Licensed under CC BY-NC-SA 4.0 (non-commercial use only).
3
+
4
+ """
5
+ Script to generate image pairs for a given scene reproducing poses provided in a metadata file.
6
+ """
7
+ import os
8
+ from datasets.habitat_sim.multiview_habitat_sim_generator import MultiviewHabitatSimGenerator
9
+ from datasets.habitat_sim.paths import SCENES_DATASET
10
+ import argparse
11
+ import quaternion
12
+ import PIL.Image
13
+ import cv2
14
+ import json
15
+ from tqdm import tqdm
16
+
17
+ def generate_multiview_images_from_metadata(metadata_filename,
18
+ output_dir,
19
+ overload_params = dict(),
20
+ scene_datasets_paths=None,
21
+ exist_ok=False):
22
+ """
23
+ Generate images from a metadata file for reproducibility purposes.
24
+ """
25
+ # Reorder paths by decreasing label length, to avoid collisions when testing if a string by such label
26
+ if scene_datasets_paths is not None:
27
+ scene_datasets_paths = dict(sorted(scene_datasets_paths.items(), key= lambda x: len(x[0]), reverse=True))
28
+
29
+ with open(metadata_filename, 'r') as f:
30
+ input_metadata = json.load(f)
31
+ metadata = dict()
32
+ for key, value in input_metadata.items():
33
+ # Optionally replace some paths
34
+ if key in ("scene_dataset_config_file", "scene", "navmesh") and value != "":
35
+ if scene_datasets_paths is not None:
36
+ for dataset_label, dataset_path in scene_datasets_paths.items():
37
+ if value.startswith(dataset_label):
38
+ value = os.path.normpath(os.path.join(dataset_path, os.path.relpath(value, dataset_label)))
39
+ break
40
+ metadata[key] = value
41
+
42
+ # Overload some parameters
43
+ for key, value in overload_params.items():
44
+ metadata[key] = value
45
+
46
+ generation_entries = dict([(key, value) for key, value in metadata.items() if not (key in ('multiviews', 'output_dir', 'generate_depth'))])
47
+ generate_depth = metadata["generate_depth"]
48
+
49
+ os.makedirs(output_dir, exist_ok=exist_ok)
50
+
51
+ generator = MultiviewHabitatSimGenerator(**generation_entries)
52
+
53
+ # Generate views
54
+ for idx_label, data in tqdm(metadata['multiviews'].items()):
55
+ positions = data["positions"]
56
+ orientations = data["orientations"]
57
+ n = len(positions)
58
+ for oidx in range(n):
59
+ observation = generator.render_viewpoint(positions[oidx], quaternion.from_float_array(orientations[oidx]))
60
+ observation_label = f"{oidx + 1}" # Leonid is indexing starting from 1
61
+ # Color image saved using PIL
62
+ img = PIL.Image.fromarray(observation['color'][:,:,:3])
63
+ filename = os.path.join(output_dir, f"{idx_label}_{observation_label}.jpeg")
64
+ img.save(filename)
65
+ if generate_depth:
66
+ # Depth image as EXR file
67
+ filename = os.path.join(output_dir, f"{idx_label}_{observation_label}_depth.exr")
68
+ cv2.imwrite(filename, observation['depth'], [cv2.IMWRITE_EXR_TYPE, cv2.IMWRITE_EXR_TYPE_HALF])
69
+ # Camera parameters
70
+ camera_params = dict([(key, observation[key].tolist()) for key in ("camera_intrinsics", "R_cam2world", "t_cam2world")])
71
+ filename = os.path.join(output_dir, f"{idx_label}_{observation_label}_camera_params.json")
72
+ with open(filename, "w") as f:
73
+ json.dump(camera_params, f)
74
+ # Save metadata
75
+ with open(os.path.join(output_dir, "metadata.json"), "w") as f:
76
+ json.dump(metadata, f)
77
+
78
+ generator.close()
79
+
80
+ if __name__ == "__main__":
81
+ parser = argparse.ArgumentParser()
82
+ parser.add_argument("--metadata_filename", required=True)
83
+ parser.add_argument("--output_dir", required=True)
84
+ args = parser.parse_args()
85
+
86
+ generate_multiview_images_from_metadata(metadata_filename=args.metadata_filename,
87
+ output_dir=args.output_dir,
88
+ scene_datasets_paths=SCENES_DATASET,
89
+ overload_params=dict(),
90
+ exist_ok=True)
91
+
92
+
demo/dust3r/croco/datasets/habitat_sim/generate_from_metadata_files.py ADDED
@@ -0,0 +1,27 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Copyright (C) 2022-present Naver Corporation. All rights reserved.
2
+ # Licensed under CC BY-NC-SA 4.0 (non-commercial use only).
3
+
4
+ """
5
+ Script generating commandlines to generate image pairs from metadata files.
6
+ """
7
+ import os
8
+ import glob
9
+ from tqdm import tqdm
10
+ import argparse
11
+
12
+ if __name__ == "__main__":
13
+ parser = argparse.ArgumentParser()
14
+ parser.add_argument("--input_dir", required=True)
15
+ parser.add_argument("--output_dir", required=True)
16
+ parser.add_argument("--prefix", default="", help="Commanline prefix, useful e.g. to setup environment.")
17
+ args = parser.parse_args()
18
+
19
+ input_metadata_filenames = glob.iglob(f"{args.input_dir}/**/metadata.json", recursive=True)
20
+
21
+ for metadata_filename in tqdm(input_metadata_filenames):
22
+ output_dir = os.path.join(args.output_dir, os.path.relpath(os.path.dirname(metadata_filename), args.input_dir))
23
+ # Do not process the scene if the metadata file already exists
24
+ if os.path.exists(os.path.join(output_dir, "metadata.json")):
25
+ continue
26
+ commandline = f"{args.prefix}python datasets/habitat_sim/generate_from_metadata.py --metadata_filename={metadata_filename} --output_dir={output_dir}"
27
+ print(commandline)
demo/dust3r/croco/datasets/habitat_sim/generate_multiview_images.py ADDED
@@ -0,0 +1,177 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Copyright (C) 2022-present Naver Corporation. All rights reserved.
2
+ # Licensed under CC BY-NC-SA 4.0 (non-commercial use only).
3
+
4
+ import os
5
+ from tqdm import tqdm
6
+ import argparse
7
+ import PIL.Image
8
+ import numpy as np
9
+ import json
10
+ from datasets.habitat_sim.multiview_habitat_sim_generator import MultiviewHabitatSimGenerator, NoNaviguableSpaceError
11
+ from datasets.habitat_sim.paths import list_scenes_available
12
+ import cv2
13
+ import quaternion
14
+ import shutil
15
+
16
+ def generate_multiview_images_for_scene(scene_dataset_config_file,
17
+ scene,
18
+ navmesh,
19
+ output_dir,
20
+ views_count,
21
+ size,
22
+ exist_ok=False,
23
+ generate_depth=False,
24
+ **kwargs):
25
+ """
26
+ Generate tuples of overlapping views for a given scene.
27
+ generate_depth: generate depth images and camera parameters.
28
+ """
29
+ if os.path.exists(output_dir) and not exist_ok:
30
+ print(f"Scene {scene}: data already generated. Ignoring generation.")
31
+ return
32
+ try:
33
+ print(f"Scene {scene}: {size} multiview acquisitions to generate...")
34
+ os.makedirs(output_dir, exist_ok=exist_ok)
35
+
36
+ metadata_filename = os.path.join(output_dir, "metadata.json")
37
+
38
+ metadata_template = dict(scene_dataset_config_file=scene_dataset_config_file,
39
+ scene=scene,
40
+ navmesh=navmesh,
41
+ views_count=views_count,
42
+ size=size,
43
+ generate_depth=generate_depth,
44
+ **kwargs)
45
+ metadata_template["multiviews"] = dict()
46
+
47
+ if os.path.exists(metadata_filename):
48
+ print("Metadata file already exists:", metadata_filename)
49
+ print("Loading already generated metadata file...")
50
+ with open(metadata_filename, "r") as f:
51
+ metadata = json.load(f)
52
+
53
+ for key in metadata_template.keys():
54
+ if key != "multiviews":
55
+ assert metadata_template[key] == metadata[key], f"existing file is inconsistent with the input parameters:\nKey: {key}\nmetadata: {metadata[key]}\ntemplate: {metadata_template[key]}."
56
+ else:
57
+ print("No temporary file found. Starting generation from scratch...")
58
+ metadata = metadata_template
59
+
60
+ starting_id = len(metadata["multiviews"])
61
+ print(f"Starting generation from index {starting_id}/{size}...")
62
+ if starting_id >= size:
63
+ print("Generation already done.")
64
+ return
65
+
66
+ generator = MultiviewHabitatSimGenerator(scene_dataset_config_file=scene_dataset_config_file,
67
+ scene=scene,
68
+ navmesh=navmesh,
69
+ views_count = views_count,
70
+ size = size,
71
+ **kwargs)
72
+
73
+ for idx in tqdm(range(starting_id, size)):
74
+ # Generate / re-generate the observations
75
+ try:
76
+ data = generator[idx]
77
+ observations = data["observations"]
78
+ positions = data["positions"]
79
+ orientations = data["orientations"]
80
+
81
+ idx_label = f"{idx:08}"
82
+ for oidx, observation in enumerate(observations):
83
+ observation_label = f"{oidx + 1}" # Leonid is indexing starting from 1
84
+ # Color image saved using PIL
85
+ img = PIL.Image.fromarray(observation['color'][:,:,:3])
86
+ filename = os.path.join(output_dir, f"{idx_label}_{observation_label}.jpeg")
87
+ img.save(filename)
88
+ if generate_depth:
89
+ # Depth image as EXR file
90
+ filename = os.path.join(output_dir, f"{idx_label}_{observation_label}_depth.exr")
91
+ cv2.imwrite(filename, observation['depth'], [cv2.IMWRITE_EXR_TYPE, cv2.IMWRITE_EXR_TYPE_HALF])
92
+ # Camera parameters
93
+ camera_params = dict([(key, observation[key].tolist()) for key in ("camera_intrinsics", "R_cam2world", "t_cam2world")])
94
+ filename = os.path.join(output_dir, f"{idx_label}_{observation_label}_camera_params.json")
95
+ with open(filename, "w") as f:
96
+ json.dump(camera_params, f)
97
+ metadata["multiviews"][idx_label] = {"positions": positions.tolist(),
98
+ "orientations": orientations.tolist(),
99
+ "covisibility_ratios": data["covisibility_ratios"].tolist(),
100
+ "valid_fractions": data["valid_fractions"].tolist(),
101
+ "pairwise_visibility_ratios": data["pairwise_visibility_ratios"].tolist()}
102
+ except RecursionError:
103
+ print("Recursion error: unable to sample observations for this scene. We will stop there.")
104
+ break
105
+
106
+ # Regularly save a temporary metadata file, in case we need to restart the generation
107
+ if idx % 10 == 0:
108
+ with open(metadata_filename, "w") as f:
109
+ json.dump(metadata, f)
110
+
111
+ # Save metadata
112
+ with open(metadata_filename, "w") as f:
113
+ json.dump(metadata, f)
114
+
115
+ generator.close()
116
+ except NoNaviguableSpaceError:
117
+ pass
118
+
119
+ def create_commandline(scene_data, generate_depth, exist_ok=False):
120
+ """
121
+ Create a commandline string to generate a scene.
122
+ """
123
+ def my_formatting(val):
124
+ if val is None or val == "":
125
+ return '""'
126
+ else:
127
+ return val
128
+ commandline = f"""python {__file__} --scene {my_formatting(scene_data.scene)}
129
+ --scene_dataset_config_file {my_formatting(scene_data.scene_dataset_config_file)}
130
+ --navmesh {my_formatting(scene_data.navmesh)}
131
+ --output_dir {my_formatting(scene_data.output_dir)}
132
+ --generate_depth {int(generate_depth)}
133
+ --exist_ok {int(exist_ok)}
134
+ """
135
+ commandline = " ".join(commandline.split())
136
+ return commandline
137
+
138
+ if __name__ == "__main__":
139
+ os.umask(2)
140
+
141
+ parser = argparse.ArgumentParser(description="""Example of use -- listing commands to generate data for scenes available:
142
+ > python datasets/habitat_sim/generate_multiview_habitat_images.py --list_commands
143
+ """)
144
+
145
+ parser.add_argument("--output_dir", type=str, required=True)
146
+ parser.add_argument("--list_commands", action='store_true', help="list commandlines to run if true")
147
+ parser.add_argument("--scene", type=str, default="")
148
+ parser.add_argument("--scene_dataset_config_file", type=str, default="")
149
+ parser.add_argument("--navmesh", type=str, default="")
150
+
151
+ parser.add_argument("--generate_depth", type=int, default=1)
152
+ parser.add_argument("--exist_ok", type=int, default=0)
153
+
154
+ kwargs = dict(resolution=(256,256), hfov=60, views_count = 2, size=1000)
155
+
156
+ args = parser.parse_args()
157
+ generate_depth=bool(args.generate_depth)
158
+ exist_ok = bool(args.exist_ok)
159
+
160
+ if args.list_commands:
161
+ # Listing scenes available...
162
+ scenes_data = list_scenes_available(base_output_dir=args.output_dir)
163
+
164
+ for scene_data in scenes_data:
165
+ print(create_commandline(scene_data, generate_depth=generate_depth, exist_ok=exist_ok))
166
+ else:
167
+ if args.scene == "" or args.output_dir == "":
168
+ print("Missing scene or output dir argument!")
169
+ print(parser.format_help())
170
+ else:
171
+ generate_multiview_images_for_scene(scene=args.scene,
172
+ scene_dataset_config_file = args.scene_dataset_config_file,
173
+ navmesh = args.navmesh,
174
+ output_dir = args.output_dir,
175
+ exist_ok=exist_ok,
176
+ generate_depth=generate_depth,
177
+ **kwargs)
demo/dust3r/croco/datasets/habitat_sim/multiview_habitat_sim_generator.py ADDED
@@ -0,0 +1,390 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Copyright (C) 2022-present Naver Corporation. All rights reserved.
2
+ # Licensed under CC BY-NC-SA 4.0 (non-commercial use only).
3
+
4
+ import os
5
+ import numpy as np
6
+ import quaternion
7
+ import habitat_sim
8
+ import json
9
+ from sklearn.neighbors import NearestNeighbors
10
+ import cv2
11
+
12
+ # OpenCV to habitat camera convention transformation
13
+ R_OPENCV2HABITAT = np.stack((habitat_sim.geo.RIGHT, -habitat_sim.geo.UP, habitat_sim.geo.FRONT), axis=0)
14
+ R_HABITAT2OPENCV = R_OPENCV2HABITAT.T
15
+ DEG2RAD = np.pi / 180
16
+
17
+ def compute_camera_intrinsics(height, width, hfov):
18
+ f = width/2 / np.tan(hfov/2 * np.pi/180)
19
+ cu, cv = width/2, height/2
20
+ return f, cu, cv
21
+
22
+ def compute_camera_pose_opencv_convention(camera_position, camera_orientation):
23
+ R_cam2world = quaternion.as_rotation_matrix(camera_orientation) @ R_OPENCV2HABITAT
24
+ t_cam2world = np.asarray(camera_position)
25
+ return R_cam2world, t_cam2world
26
+
27
+ def compute_pointmap(depthmap, hfov):
28
+ """ Compute a HxWx3 pointmap in camera frame from a HxW depth map."""
29
+ height, width = depthmap.shape
30
+ f, cu, cv = compute_camera_intrinsics(height, width, hfov)
31
+ # Cast depth map to point
32
+ z_cam = depthmap
33
+ u, v = np.meshgrid(range(width), range(height))
34
+ x_cam = (u - cu) / f * z_cam
35
+ y_cam = (v - cv) / f * z_cam
36
+ X_cam = np.stack((x_cam, y_cam, z_cam), axis=-1)
37
+ return X_cam
38
+
39
+ def compute_pointcloud(depthmap, hfov, camera_position, camera_rotation):
40
+ """Return a 3D point cloud corresponding to valid pixels of the depth map"""
41
+ R_cam2world, t_cam2world = compute_camera_pose_opencv_convention(camera_position, camera_rotation)
42
+
43
+ X_cam = compute_pointmap(depthmap=depthmap, hfov=hfov)
44
+ valid_mask = (X_cam[:,:,2] != 0.0)
45
+
46
+ X_cam = X_cam.reshape(-1, 3)[valid_mask.flatten()]
47
+ X_world = X_cam @ R_cam2world.T + t_cam2world.reshape(1, 3)
48
+ return X_world
49
+
50
+ def compute_pointcloud_overlaps_scikit(pointcloud1, pointcloud2, distance_threshold, compute_symmetric=False):
51
+ """
52
+ Compute 'overlapping' metrics based on a distance threshold between two point clouds.
53
+ """
54
+ nbrs = NearestNeighbors(n_neighbors=1, algorithm = 'kd_tree').fit(pointcloud2)
55
+ distances, indices = nbrs.kneighbors(pointcloud1)
56
+ intersection1 = np.count_nonzero(distances.flatten() < distance_threshold)
57
+
58
+ data = {"intersection1": intersection1,
59
+ "size1": len(pointcloud1)}
60
+ if compute_symmetric:
61
+ nbrs = NearestNeighbors(n_neighbors=1, algorithm = 'kd_tree').fit(pointcloud1)
62
+ distances, indices = nbrs.kneighbors(pointcloud2)
63
+ intersection2 = np.count_nonzero(distances.flatten() < distance_threshold)
64
+ data["intersection2"] = intersection2
65
+ data["size2"] = len(pointcloud2)
66
+
67
+ return data
68
+
69
+ def _append_camera_parameters(observation, hfov, camera_location, camera_rotation):
70
+ """
71
+ Add camera parameters to the observation dictionnary produced by Habitat-Sim
72
+ In-place modifications.
73
+ """
74
+ R_cam2world, t_cam2world = compute_camera_pose_opencv_convention(camera_location, camera_rotation)
75
+ height, width = observation['depth'].shape
76
+ f, cu, cv = compute_camera_intrinsics(height, width, hfov)
77
+ K = np.asarray([[f, 0, cu],
78
+ [0, f, cv],
79
+ [0, 0, 1.0]])
80
+ observation["camera_intrinsics"] = K
81
+ observation["t_cam2world"] = t_cam2world
82
+ observation["R_cam2world"] = R_cam2world
83
+
84
+ def look_at(eye, center, up, return_cam2world=True):
85
+ """
86
+ Return camera pose looking at a given center point.
87
+ Analogous of gluLookAt function, using OpenCV camera convention.
88
+ """
89
+ z = center - eye
90
+ z /= np.linalg.norm(z, axis=-1, keepdims=True)
91
+ y = -up
92
+ y = y - np.sum(y * z, axis=-1, keepdims=True) * z
93
+ y /= np.linalg.norm(y, axis=-1, keepdims=True)
94
+ x = np.cross(y, z, axis=-1)
95
+
96
+ if return_cam2world:
97
+ R = np.stack((x, y, z), axis=-1)
98
+ t = eye
99
+ else:
100
+ # World to camera transformation
101
+ # Transposed matrix
102
+ R = np.stack((x, y, z), axis=-2)
103
+ t = - np.einsum('...ij, ...j', R, eye)
104
+ return R, t
105
+
106
+ def look_at_for_habitat(eye, center, up, return_cam2world=True):
107
+ R, t = look_at(eye, center, up)
108
+ orientation = quaternion.from_rotation_matrix(R @ R_OPENCV2HABITAT.T)
109
+ return orientation, t
110
+
111
+ def generate_orientation_noise(pan_range, tilt_range, roll_range):
112
+ return (quaternion.from_rotation_vector(np.random.uniform(*pan_range) * DEG2RAD * habitat_sim.geo.UP)
113
+ * quaternion.from_rotation_vector(np.random.uniform(*tilt_range) * DEG2RAD * habitat_sim.geo.RIGHT)
114
+ * quaternion.from_rotation_vector(np.random.uniform(*roll_range) * DEG2RAD * habitat_sim.geo.FRONT))
115
+
116
+
117
+ class NoNaviguableSpaceError(RuntimeError):
118
+ def __init__(self, *args):
119
+ super().__init__(*args)
120
+
121
+ class MultiviewHabitatSimGenerator:
122
+ def __init__(self,
123
+ scene,
124
+ navmesh,
125
+ scene_dataset_config_file,
126
+ resolution = (240, 320),
127
+ views_count=2,
128
+ hfov = 60,
129
+ gpu_id = 0,
130
+ size = 10000,
131
+ minimum_covisibility = 0.5,
132
+ transform = None):
133
+ self.scene = scene
134
+ self.navmesh = navmesh
135
+ self.scene_dataset_config_file = scene_dataset_config_file
136
+ self.resolution = resolution
137
+ self.views_count = views_count
138
+ assert(self.views_count >= 1)
139
+ self.hfov = hfov
140
+ self.gpu_id = gpu_id
141
+ self.size = size
142
+ self.transform = transform
143
+
144
+ # Noise added to camera orientation
145
+ self.pan_range = (-3, 3)
146
+ self.tilt_range = (-10, 10)
147
+ self.roll_range = (-5, 5)
148
+
149
+ # Height range to sample cameras
150
+ self.height_range = (1.2, 1.8)
151
+
152
+ # Random steps between the camera views
153
+ self.random_steps_count = 5
154
+ self.random_step_variance = 2.0
155
+
156
+ # Minimum fraction of the scene which should be valid (well defined depth)
157
+ self.minimum_valid_fraction = 0.7
158
+
159
+ # Distance threshold to see to select pairs
160
+ self.distance_threshold = 0.05
161
+ # Minimum IoU of a view point cloud with respect to the reference view to be kept.
162
+ self.minimum_covisibility = minimum_covisibility
163
+
164
+ # Maximum number of retries.
165
+ self.max_attempts_count = 100
166
+
167
+ self.seed = None
168
+ self._lazy_initialization()
169
+
170
+ def _lazy_initialization(self):
171
+ # Lazy random seeding and instantiation of the simulator to deal with multiprocessing properly
172
+ if self.seed == None:
173
+ # Re-seed numpy generator
174
+ np.random.seed()
175
+ self.seed = np.random.randint(2**32-1)
176
+ sim_cfg = habitat_sim.SimulatorConfiguration()
177
+ sim_cfg.scene_id = self.scene
178
+ if self.scene_dataset_config_file is not None and self.scene_dataset_config_file != "":
179
+ sim_cfg.scene_dataset_config_file = self.scene_dataset_config_file
180
+ sim_cfg.random_seed = self.seed
181
+ sim_cfg.load_semantic_mesh = False
182
+ sim_cfg.gpu_device_id = self.gpu_id
183
+
184
+ depth_sensor_spec = habitat_sim.CameraSensorSpec()
185
+ depth_sensor_spec.uuid = "depth"
186
+ depth_sensor_spec.sensor_type = habitat_sim.SensorType.DEPTH
187
+ depth_sensor_spec.resolution = self.resolution
188
+ depth_sensor_spec.hfov = self.hfov
189
+ depth_sensor_spec.position = [0.0, 0.0, 0]
190
+ depth_sensor_spec.orientation
191
+
192
+ rgb_sensor_spec = habitat_sim.CameraSensorSpec()
193
+ rgb_sensor_spec.uuid = "color"
194
+ rgb_sensor_spec.sensor_type = habitat_sim.SensorType.COLOR
195
+ rgb_sensor_spec.resolution = self.resolution
196
+ rgb_sensor_spec.hfov = self.hfov
197
+ rgb_sensor_spec.position = [0.0, 0.0, 0]
198
+ agent_cfg = habitat_sim.agent.AgentConfiguration(sensor_specifications=[rgb_sensor_spec, depth_sensor_spec])
199
+
200
+ cfg = habitat_sim.Configuration(sim_cfg, [agent_cfg])
201
+ self.sim = habitat_sim.Simulator(cfg)
202
+ if self.navmesh is not None and self.navmesh != "":
203
+ # Use pre-computed navmesh when available (usually better than those generated automatically)
204
+ self.sim.pathfinder.load_nav_mesh(self.navmesh)
205
+
206
+ if not self.sim.pathfinder.is_loaded:
207
+ # Try to compute a navmesh
208
+ navmesh_settings = habitat_sim.NavMeshSettings()
209
+ navmesh_settings.set_defaults()
210
+ self.sim.recompute_navmesh(self.sim.pathfinder, navmesh_settings, True)
211
+
212
+ # Ensure that the navmesh is not empty
213
+ if not self.sim.pathfinder.is_loaded:
214
+ raise NoNaviguableSpaceError(f"No naviguable location (scene: {self.scene} -- navmesh: {self.navmesh})")
215
+
216
+ self.agent = self.sim.initialize_agent(agent_id=0)
217
+
218
+ def close(self):
219
+ self.sim.close()
220
+
221
+ def __del__(self):
222
+ self.sim.close()
223
+
224
+ def __len__(self):
225
+ return self.size
226
+
227
+ def sample_random_viewpoint(self):
228
+ """ Sample a random viewpoint using the navmesh """
229
+ nav_point = self.sim.pathfinder.get_random_navigable_point()
230
+
231
+ # Sample a random viewpoint height
232
+ viewpoint_height = np.random.uniform(*self.height_range)
233
+ viewpoint_position = nav_point + viewpoint_height * habitat_sim.geo.UP
234
+ viewpoint_orientation = quaternion.from_rotation_vector(np.random.uniform(0, 2 * np.pi) * habitat_sim.geo.UP) * generate_orientation_noise(self.pan_range, self.tilt_range, self.roll_range)
235
+ return viewpoint_position, viewpoint_orientation, nav_point
236
+
237
+ def sample_other_random_viewpoint(self, observed_point, nav_point):
238
+ """ Sample a random viewpoint close to an existing one, using the navmesh and a reference observed point."""
239
+ other_nav_point = nav_point
240
+
241
+ walk_directions = self.random_step_variance * np.asarray([1,0,1])
242
+ for i in range(self.random_steps_count):
243
+ temp = self.sim.pathfinder.snap_point(other_nav_point + walk_directions * np.random.normal(size=3))
244
+ # Snapping may return nan when it fails
245
+ if not np.isnan(temp[0]):
246
+ other_nav_point = temp
247
+
248
+ other_viewpoint_height = np.random.uniform(*self.height_range)
249
+ other_viewpoint_position = other_nav_point + other_viewpoint_height * habitat_sim.geo.UP
250
+
251
+ # Set viewing direction towards the central point
252
+ rotation, position = look_at_for_habitat(eye=other_viewpoint_position, center=observed_point, up=habitat_sim.geo.UP, return_cam2world=True)
253
+ rotation = rotation * generate_orientation_noise(self.pan_range, self.tilt_range, self.roll_range)
254
+ return position, rotation, other_nav_point
255
+
256
+ def is_other_pointcloud_overlapping(self, ref_pointcloud, other_pointcloud):
257
+ """ Check if a viewpoint is valid and overlaps significantly with a reference one. """
258
+ # Observation
259
+ pixels_count = self.resolution[0] * self.resolution[1]
260
+ valid_fraction = len(other_pointcloud) / pixels_count
261
+ assert valid_fraction <= 1.0 and valid_fraction >= 0.0
262
+ overlap = compute_pointcloud_overlaps_scikit(ref_pointcloud, other_pointcloud, self.distance_threshold, compute_symmetric=True)
263
+ covisibility = min(overlap["intersection1"] / pixels_count, overlap["intersection2"] / pixels_count)
264
+ is_valid = (valid_fraction >= self.minimum_valid_fraction) and (covisibility >= self.minimum_covisibility)
265
+ return is_valid, valid_fraction, covisibility
266
+
267
+ def is_other_viewpoint_overlapping(self, ref_pointcloud, observation, position, rotation):
268
+ """ Check if a viewpoint is valid and overlaps significantly with a reference one. """
269
+ # Observation
270
+ other_pointcloud = compute_pointcloud(observation['depth'], self.hfov, position, rotation)
271
+ return self.is_other_pointcloud_overlapping(ref_pointcloud, other_pointcloud)
272
+
273
+ def render_viewpoint(self, viewpoint_position, viewpoint_orientation):
274
+ agent_state = habitat_sim.AgentState()
275
+ agent_state.position = viewpoint_position
276
+ agent_state.rotation = viewpoint_orientation
277
+ self.agent.set_state(agent_state)
278
+ viewpoint_observations = self.sim.get_sensor_observations(agent_ids=0)
279
+ _append_camera_parameters(viewpoint_observations, self.hfov, viewpoint_position, viewpoint_orientation)
280
+ return viewpoint_observations
281
+
282
+ def __getitem__(self, useless_idx):
283
+ ref_position, ref_orientation, nav_point = self.sample_random_viewpoint()
284
+ ref_observations = self.render_viewpoint(ref_position, ref_orientation)
285
+ # Extract point cloud
286
+ ref_pointcloud = compute_pointcloud(depthmap=ref_observations['depth'], hfov=self.hfov,
287
+ camera_position=ref_position, camera_rotation=ref_orientation)
288
+
289
+ pixels_count = self.resolution[0] * self.resolution[1]
290
+ ref_valid_fraction = len(ref_pointcloud) / pixels_count
291
+ assert ref_valid_fraction <= 1.0 and ref_valid_fraction >= 0.0
292
+ if ref_valid_fraction < self.minimum_valid_fraction:
293
+ # This should produce a recursion error at some point when something is very wrong.
294
+ return self[0]
295
+ # Pick an reference observed point in the point cloud
296
+ observed_point = np.mean(ref_pointcloud, axis=0)
297
+
298
+ # Add the first image as reference
299
+ viewpoints_observations = [ref_observations]
300
+ viewpoints_covisibility = [ref_valid_fraction]
301
+ viewpoints_positions = [ref_position]
302
+ viewpoints_orientations = [quaternion.as_float_array(ref_orientation)]
303
+ viewpoints_clouds = [ref_pointcloud]
304
+ viewpoints_valid_fractions = [ref_valid_fraction]
305
+
306
+ for _ in range(self.views_count - 1):
307
+ # Generate an other viewpoint using some dummy random walk
308
+ successful_sampling = False
309
+ for sampling_attempt in range(self.max_attempts_count):
310
+ position, rotation, _ = self.sample_other_random_viewpoint(observed_point, nav_point)
311
+ # Observation
312
+ other_viewpoint_observations = self.render_viewpoint(position, rotation)
313
+ other_pointcloud = compute_pointcloud(other_viewpoint_observations['depth'], self.hfov, position, rotation)
314
+
315
+ is_valid, valid_fraction, covisibility = self.is_other_pointcloud_overlapping(ref_pointcloud, other_pointcloud)
316
+ if is_valid:
317
+ successful_sampling = True
318
+ break
319
+ if not successful_sampling:
320
+ print("WARNING: Maximum number of attempts reached.")
321
+ # Dirty hack, try using a novel original viewpoint
322
+ return self[0]
323
+ viewpoints_observations.append(other_viewpoint_observations)
324
+ viewpoints_covisibility.append(covisibility)
325
+ viewpoints_positions.append(position)
326
+ viewpoints_orientations.append(quaternion.as_float_array(rotation)) # WXYZ convention for the quaternion encoding.
327
+ viewpoints_clouds.append(other_pointcloud)
328
+ viewpoints_valid_fractions.append(valid_fraction)
329
+
330
+ # Estimate relations between all pairs of images
331
+ pairwise_visibility_ratios = np.ones((len(viewpoints_observations), len(viewpoints_observations)))
332
+ for i in range(len(viewpoints_observations)):
333
+ pairwise_visibility_ratios[i,i] = viewpoints_valid_fractions[i]
334
+ for j in range(i+1, len(viewpoints_observations)):
335
+ overlap = compute_pointcloud_overlaps_scikit(viewpoints_clouds[i], viewpoints_clouds[j], self.distance_threshold, compute_symmetric=True)
336
+ pairwise_visibility_ratios[i,j] = overlap['intersection1'] / pixels_count
337
+ pairwise_visibility_ratios[j,i] = overlap['intersection2'] / pixels_count
338
+
339
+ # IoU is relative to the image 0
340
+ data = {"observations": viewpoints_observations,
341
+ "positions": np.asarray(viewpoints_positions),
342
+ "orientations": np.asarray(viewpoints_orientations),
343
+ "covisibility_ratios": np.asarray(viewpoints_covisibility),
344
+ "valid_fractions": np.asarray(viewpoints_valid_fractions, dtype=float),
345
+ "pairwise_visibility_ratios": np.asarray(pairwise_visibility_ratios, dtype=float),
346
+ }
347
+
348
+ if self.transform is not None:
349
+ data = self.transform(data)
350
+ return data
351
+
352
+ def generate_random_spiral_trajectory(self, images_count = 100, max_radius=0.5, half_turns=5, use_constant_orientation=False):
353
+ """
354
+ Return a list of images corresponding to a spiral trajectory from a random starting point.
355
+ Useful to generate nice visualisations.
356
+ Use an even number of half turns to get a nice "C1-continuous" loop effect
357
+ """
358
+ ref_position, ref_orientation, navpoint = self.sample_random_viewpoint()
359
+ ref_observations = self.render_viewpoint(ref_position, ref_orientation)
360
+ ref_pointcloud = compute_pointcloud(depthmap=ref_observations['depth'], hfov=self.hfov,
361
+ camera_position=ref_position, camera_rotation=ref_orientation)
362
+ pixels_count = self.resolution[0] * self.resolution[1]
363
+ if len(ref_pointcloud) / pixels_count < self.minimum_valid_fraction:
364
+ # Dirty hack: ensure that the valid part of the image is significant
365
+ return self.generate_random_spiral_trajectory(images_count, max_radius, half_turns, use_constant_orientation)
366
+
367
+ # Pick an observed point in the point cloud
368
+ observed_point = np.mean(ref_pointcloud, axis=0)
369
+ ref_R, ref_t = compute_camera_pose_opencv_convention(ref_position, ref_orientation)
370
+
371
+ images = []
372
+ is_valid = []
373
+ # Spiral trajectory, use_constant orientation
374
+ for i, alpha in enumerate(np.linspace(0, 1, images_count)):
375
+ r = max_radius * np.abs(np.sin(alpha * np.pi)) # Increase then decrease the radius
376
+ theta = alpha * half_turns * np.pi
377
+ x = r * np.cos(theta)
378
+ y = r * np.sin(theta)
379
+ z = 0.0
380
+ position = ref_position + (ref_R @ np.asarray([x, y, z]).reshape(3,1)).flatten()
381
+ if use_constant_orientation:
382
+ orientation = ref_orientation
383
+ else:
384
+ # trajectory looking at a mean point in front of the ref observation
385
+ orientation, position = look_at_for_habitat(eye=position, center=observed_point, up=habitat_sim.geo.UP)
386
+ observations = self.render_viewpoint(position, orientation)
387
+ images.append(observations['color'][...,:3])
388
+ _is_valid, valid_fraction, iou = self.is_other_viewpoint_overlapping(ref_pointcloud, observations, position, orientation)
389
+ is_valid.append(_is_valid)
390
+ return images, np.all(is_valid)
demo/dust3r/croco/datasets/habitat_sim/pack_metadata_files.py ADDED
@@ -0,0 +1,69 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Copyright (C) 2022-present Naver Corporation. All rights reserved.
2
+ # Licensed under CC BY-NC-SA 4.0 (non-commercial use only).
3
+ """
4
+ Utility script to pack metadata files of the dataset in order to be able to re-generate it elsewhere.
5
+ """
6
+ import os
7
+ import glob
8
+ from tqdm import tqdm
9
+ import shutil
10
+ import json
11
+ from datasets.habitat_sim.paths import *
12
+ import argparse
13
+ import collections
14
+
15
+ if __name__ == "__main__":
16
+ parser = argparse.ArgumentParser()
17
+ parser.add_argument("input_dir")
18
+ parser.add_argument("output_dir")
19
+ args = parser.parse_args()
20
+
21
+ input_dirname = args.input_dir
22
+ output_dirname = args.output_dir
23
+
24
+ input_metadata_filenames = glob.iglob(f"{input_dirname}/**/metadata.json", recursive=True)
25
+
26
+ images_count = collections.defaultdict(lambda : 0)
27
+
28
+ os.makedirs(output_dirname)
29
+ for input_filename in tqdm(input_metadata_filenames):
30
+ # Ignore empty files
31
+ with open(input_filename, "r") as f:
32
+ original_metadata = json.load(f)
33
+ if "multiviews" not in original_metadata or len(original_metadata["multiviews"]) == 0:
34
+ print("No views in", input_filename)
35
+ continue
36
+
37
+ relpath = os.path.relpath(input_filename, input_dirname)
38
+ print(relpath)
39
+
40
+ # Copy metadata, while replacing scene paths by generic keys depending on the dataset, for portability.
41
+ # Data paths are sorted by decreasing length to avoid potential bugs due to paths starting by the same string pattern.
42
+ scenes_dataset_paths = dict(sorted(SCENES_DATASET.items(), key=lambda x: len(x[1]), reverse=True))
43
+ metadata = dict()
44
+ for key, value in original_metadata.items():
45
+ if key in ("scene_dataset_config_file", "scene", "navmesh") and value != "":
46
+ known_path = False
47
+ for dataset, dataset_path in scenes_dataset_paths.items():
48
+ if value.startswith(dataset_path):
49
+ value = os.path.join(dataset, os.path.relpath(value, dataset_path))
50
+ known_path = True
51
+ break
52
+ if not known_path:
53
+ raise KeyError("Unknown path:" + value)
54
+ metadata[key] = value
55
+
56
+ # Compile some general statistics while packing data
57
+ scene_split = metadata["scene"].split("/")
58
+ upper_level = "/".join(scene_split[:2]) if scene_split[0] == "hm3d" else scene_split[0]
59
+ images_count[upper_level] += len(metadata["multiviews"])
60
+
61
+ output_filename = os.path.join(output_dirname, relpath)
62
+ os.makedirs(os.path.dirname(output_filename), exist_ok=True)
63
+ with open(output_filename, "w") as f:
64
+ json.dump(metadata, f)
65
+
66
+ # Print statistics
67
+ print("Images count:")
68
+ for upper_level, count in images_count.items():
69
+ print(f"- {upper_level}: {count}")
demo/dust3r/croco/datasets/habitat_sim/paths.py ADDED
@@ -0,0 +1,129 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Copyright (C) 2022-present Naver Corporation. All rights reserved.
2
+ # Licensed under CC BY-NC-SA 4.0 (non-commercial use only).
3
+
4
+ """
5
+ Paths to Habitat-Sim scenes
6
+ """
7
+
8
+ import os
9
+ import json
10
+ import collections
11
+ from tqdm import tqdm
12
+
13
+
14
+ # Hardcoded path to the different scene datasets
15
+ SCENES_DATASET = {
16
+ "hm3d": "./data/habitat-sim-data/scene_datasets/hm3d/",
17
+ "gibson": "./data/habitat-sim-data/scene_datasets/gibson/",
18
+ "habitat-test-scenes": "./data/habitat-sim/scene_datasets/habitat-test-scenes/",
19
+ "replica_cad_baked_lighting": "./data/habitat-sim/scene_datasets/replica_cad_baked_lighting/",
20
+ "replica_cad": "./data/habitat-sim/scene_datasets/replica_cad/",
21
+ "replica": "./data/habitat-sim/scene_datasets/ReplicaDataset/",
22
+ "scannet": "./data/habitat-sim/scene_datasets/scannet/"
23
+ }
24
+
25
+ SceneData = collections.namedtuple("SceneData", ["scene_dataset_config_file", "scene", "navmesh", "output_dir"])
26
+
27
+ def list_replicacad_scenes(base_output_dir, base_path=SCENES_DATASET["replica_cad"]):
28
+ scene_dataset_config_file = os.path.join(base_path, "replicaCAD.scene_dataset_config.json")
29
+ scenes = [f"apt_{i}" for i in range(6)] + ["empty_stage"]
30
+ navmeshes = [f"navmeshes/apt_{i}_static_furniture.navmesh" for i in range(6)] + ["empty_stage.navmesh"]
31
+ scenes_data = []
32
+ for idx in range(len(scenes)):
33
+ output_dir = os.path.join(base_output_dir, "ReplicaCAD", scenes[idx])
34
+ # Add scene
35
+ data = SceneData(scene_dataset_config_file=scene_dataset_config_file,
36
+ scene = scenes[idx] + ".scene_instance.json",
37
+ navmesh = os.path.join(base_path, navmeshes[idx]),
38
+ output_dir = output_dir)
39
+ scenes_data.append(data)
40
+ return scenes_data
41
+
42
+ def list_replica_cad_baked_lighting_scenes(base_output_dir, base_path=SCENES_DATASET["replica_cad_baked_lighting"]):
43
+ scene_dataset_config_file = os.path.join(base_path, "replicaCAD_baked.scene_dataset_config.json")
44
+ scenes = sum([[f"Baked_sc{i}_staging_{j:02}" for i in range(5)] for j in range(21)], [])
45
+ navmeshes = ""#[f"navmeshes/apt_{i}_static_furniture.navmesh" for i in range(6)] + ["empty_stage.navmesh"]
46
+ scenes_data = []
47
+ for idx in range(len(scenes)):
48
+ output_dir = os.path.join(base_output_dir, "replica_cad_baked_lighting", scenes[idx])
49
+ data = SceneData(scene_dataset_config_file=scene_dataset_config_file,
50
+ scene = scenes[idx],
51
+ navmesh = "",
52
+ output_dir = output_dir)
53
+ scenes_data.append(data)
54
+ return scenes_data
55
+
56
+ def list_replica_scenes(base_output_dir, base_path):
57
+ scenes_data = []
58
+ for scene_id in os.listdir(base_path):
59
+ scene = os.path.join(base_path, scene_id, "mesh.ply")
60
+ navmesh = os.path.join(base_path, scene_id, "habitat/mesh_preseg_semantic.navmesh") # Not sure if I should use it
61
+ scene_dataset_config_file = ""
62
+ output_dir = os.path.join(base_output_dir, scene_id)
63
+ # Add scene only if it does not exist already, or if exist_ok
64
+ data = SceneData(scene_dataset_config_file = scene_dataset_config_file,
65
+ scene = scene,
66
+ navmesh = navmesh,
67
+ output_dir = output_dir)
68
+ scenes_data.append(data)
69
+ return scenes_data
70
+
71
+
72
+ def list_scenes(base_output_dir, base_path):
73
+ """
74
+ Generic method iterating through a base_path folder to find scenes.
75
+ """
76
+ scenes_data = []
77
+ for root, dirs, files in os.walk(base_path, followlinks=True):
78
+ folder_scenes_data = []
79
+ for file in files:
80
+ name, ext = os.path.splitext(file)
81
+ if ext == ".glb":
82
+ scene = os.path.join(root, name + ".glb")
83
+ navmesh = os.path.join(root, name + ".navmesh")
84
+ if not os.path.exists(navmesh):
85
+ navmesh = ""
86
+ relpath = os.path.relpath(root, base_path)
87
+ output_dir = os.path.abspath(os.path.join(base_output_dir, relpath, name))
88
+ data = SceneData(scene_dataset_config_file="",
89
+ scene = scene,
90
+ navmesh = navmesh,
91
+ output_dir = output_dir)
92
+ folder_scenes_data.append(data)
93
+
94
+ # Specific check for HM3D:
95
+ # When two meshesxxxx.basis.glb and xxxx.glb are present, use the 'basis' version.
96
+ basis_scenes = [data.scene[:-len(".basis.glb")] for data in folder_scenes_data if data.scene.endswith(".basis.glb")]
97
+ if len(basis_scenes) != 0:
98
+ folder_scenes_data = [data for data in folder_scenes_data if not (data.scene[:-len(".glb")] in basis_scenes)]
99
+
100
+ scenes_data.extend(folder_scenes_data)
101
+ return scenes_data
102
+
103
+ def list_scenes_available(base_output_dir, scenes_dataset_paths=SCENES_DATASET):
104
+ scenes_data = []
105
+
106
+ # HM3D
107
+ for split in ("minival", "train", "val", "examples"):
108
+ scenes_data += list_scenes(base_output_dir=os.path.join(base_output_dir, f"hm3d/{split}/"),
109
+ base_path=f"{scenes_dataset_paths['hm3d']}/{split}")
110
+
111
+ # Gibson
112
+ scenes_data += list_scenes(base_output_dir=os.path.join(base_output_dir, "gibson"),
113
+ base_path=scenes_dataset_paths["gibson"])
114
+
115
+ # Habitat test scenes (just a few)
116
+ scenes_data += list_scenes(base_output_dir=os.path.join(base_output_dir, "habitat-test-scenes"),
117
+ base_path=scenes_dataset_paths["habitat-test-scenes"])
118
+
119
+ # ReplicaCAD (baked lightning)
120
+ scenes_data += list_replica_cad_baked_lighting_scenes(base_output_dir=base_output_dir)
121
+
122
+ # ScanNet
123
+ scenes_data += list_scenes(base_output_dir=os.path.join(base_output_dir, "scannet"),
124
+ base_path=scenes_dataset_paths["scannet"])
125
+
126
+ # Replica
127
+ list_replica_scenes(base_output_dir=os.path.join(base_output_dir, "replica"),
128
+ base_path=scenes_dataset_paths["replica"])
129
+ return scenes_data
demo/dust3r/croco/datasets/pairs_dataset.py ADDED
@@ -0,0 +1,109 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Copyright (C) 2022-present Naver Corporation. All rights reserved.
2
+ # Licensed under CC BY-NC-SA 4.0 (non-commercial use only).
3
+
4
+ import os
5
+ from torch.utils.data import Dataset
6
+ from PIL import Image
7
+
8
+ from datasets.transforms import get_pair_transforms
9
+
10
+ def load_image(impath):
11
+ return Image.open(impath)
12
+
13
+ def load_pairs_from_cache_file(fname, root=''):
14
+ assert os.path.isfile(fname), "cannot parse pairs from {:s}, file does not exist".format(fname)
15
+ with open(fname, 'r') as fid:
16
+ lines = fid.read().strip().splitlines()
17
+ pairs = [ (os.path.join(root,l.split()[0]), os.path.join(root,l.split()[1])) for l in lines]
18
+ return pairs
19
+
20
+ def load_pairs_from_list_file(fname, root=''):
21
+ assert os.path.isfile(fname), "cannot parse pairs from {:s}, file does not exist".format(fname)
22
+ with open(fname, 'r') as fid:
23
+ lines = fid.read().strip().splitlines()
24
+ pairs = [ (os.path.join(root,l+'_1.jpg'), os.path.join(root,l+'_2.jpg')) for l in lines if not l.startswith('#')]
25
+ return pairs
26
+
27
+
28
+ def write_cache_file(fname, pairs, root=''):
29
+ if len(root)>0:
30
+ if not root.endswith('/'): root+='/'
31
+ assert os.path.isdir(root)
32
+ s = ''
33
+ for im1, im2 in pairs:
34
+ if len(root)>0:
35
+ assert im1.startswith(root), im1
36
+ assert im2.startswith(root), im2
37
+ s += '{:s} {:s}\n'.format(im1[len(root):], im2[len(root):])
38
+ with open(fname, 'w') as fid:
39
+ fid.write(s[:-1])
40
+
41
+ def parse_and_cache_all_pairs(dname, data_dir='./data/'):
42
+ if dname=='habitat_release':
43
+ dirname = os.path.join(data_dir, 'habitat_release')
44
+ assert os.path.isdir(dirname), "cannot find folder for habitat_release pairs: "+dirname
45
+ cache_file = os.path.join(dirname, 'pairs.txt')
46
+ assert not os.path.isfile(cache_file), "cache file already exists: "+cache_file
47
+
48
+ print('Parsing pairs for dataset: '+dname)
49
+ pairs = []
50
+ for root, dirs, files in os.walk(dirname):
51
+ if 'val' in root: continue
52
+ dirs.sort()
53
+ pairs += [ (os.path.join(root,f), os.path.join(root,f[:-len('_1.jpeg')]+'_2.jpeg')) for f in sorted(files) if f.endswith('_1.jpeg')]
54
+ print('Found {:,} pairs'.format(len(pairs)))
55
+ print('Writing cache to: '+cache_file)
56
+ write_cache_file(cache_file, pairs, root=dirname)
57
+
58
+ else:
59
+ raise NotImplementedError('Unknown dataset: '+dname)
60
+
61
+ def dnames_to_image_pairs(dnames, data_dir='./data/'):
62
+ """
63
+ dnames: list of datasets with image pairs, separated by +
64
+ """
65
+ all_pairs = []
66
+ for dname in dnames.split('+'):
67
+ if dname=='habitat_release':
68
+ dirname = os.path.join(data_dir, 'habitat_release')
69
+ assert os.path.isdir(dirname), "cannot find folder for habitat_release pairs: "+dirname
70
+ cache_file = os.path.join(dirname, 'pairs.txt')
71
+ assert os.path.isfile(cache_file), "cannot find cache file for habitat_release pairs, please first create the cache file, see instructions. "+cache_file
72
+ pairs = load_pairs_from_cache_file(cache_file, root=dirname)
73
+ elif dname in ['ARKitScenes', 'MegaDepth', '3DStreetView', 'IndoorVL']:
74
+ dirname = os.path.join(data_dir, dname+'_crops')
75
+ assert os.path.isdir(dirname), "cannot find folder for {:s} pairs: {:s}".format(dname, dirname)
76
+ list_file = os.path.join(dirname, 'listing.txt')
77
+ assert os.path.isfile(list_file), "cannot find list file for {:s} pairs, see instructions. {:s}".format(dname, list_file)
78
+ pairs = load_pairs_from_list_file(list_file, root=dirname)
79
+ print(' {:s}: {:,} pairs'.format(dname, len(pairs)))
80
+ all_pairs += pairs
81
+ if '+' in dnames: print(' Total: {:,} pairs'.format(len(all_pairs)))
82
+ return all_pairs
83
+
84
+
85
+ class PairsDataset(Dataset):
86
+
87
+ def __init__(self, dnames, trfs='', totensor=True, normalize=True, data_dir='./data/'):
88
+ super().__init__()
89
+ self.image_pairs = dnames_to_image_pairs(dnames, data_dir=data_dir)
90
+ self.transforms = get_pair_transforms(transform_str=trfs, totensor=totensor, normalize=normalize)
91
+
92
+ def __len__(self):
93
+ return len(self.image_pairs)
94
+
95
+ def __getitem__(self, index):
96
+ im1path, im2path = self.image_pairs[index]
97
+ im1 = load_image(im1path)
98
+ im2 = load_image(im2path)
99
+ if self.transforms is not None: im1, im2 = self.transforms(im1, im2)
100
+ return im1, im2
101
+
102
+
103
+ if __name__=="__main__":
104
+ import argparse
105
+ parser = argparse.ArgumentParser(prog="Computing and caching list of pairs for a given dataset")
106
+ parser.add_argument('--data_dir', default='./data/', type=str, help="path where data are stored")
107
+ parser.add_argument('--dataset', default='habitat_release', type=str, help="name of the dataset")
108
+ args = parser.parse_args()
109
+ parse_and_cache_all_pairs(dname=args.dataset, data_dir=args.data_dir)
demo/dust3r/croco/datasets/transforms.py ADDED
@@ -0,0 +1,95 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Copyright (C) 2022-present Naver Corporation. All rights reserved.
2
+ # Licensed under CC BY-NC-SA 4.0 (non-commercial use only).
3
+
4
+ import torch
5
+ import torchvision.transforms
6
+ import torchvision.transforms.functional as F
7
+
8
+ # "Pair": apply a transform on a pair
9
+ # "Both": apply the exact same transform to both images
10
+
11
+ class ComposePair(torchvision.transforms.Compose):
12
+ def __call__(self, img1, img2):
13
+ for t in self.transforms:
14
+ img1, img2 = t(img1, img2)
15
+ return img1, img2
16
+
17
+ class NormalizeBoth(torchvision.transforms.Normalize):
18
+ def forward(self, img1, img2):
19
+ img1 = super().forward(img1)
20
+ img2 = super().forward(img2)
21
+ return img1, img2
22
+
23
+ class ToTensorBoth(torchvision.transforms.ToTensor):
24
+ def __call__(self, img1, img2):
25
+ img1 = super().__call__(img1)
26
+ img2 = super().__call__(img2)
27
+ return img1, img2
28
+
29
+ class RandomCropPair(torchvision.transforms.RandomCrop):
30
+ # the crop will be intentionally different for the two images with this class
31
+ def forward(self, img1, img2):
32
+ img1 = super().forward(img1)
33
+ img2 = super().forward(img2)
34
+ return img1, img2
35
+
36
+ class ColorJitterPair(torchvision.transforms.ColorJitter):
37
+ # can be symmetric (same for both images) or assymetric (different jitter params for each image) depending on assymetric_prob
38
+ def __init__(self, assymetric_prob, **kwargs):
39
+ super().__init__(**kwargs)
40
+ self.assymetric_prob = assymetric_prob
41
+ def jitter_one(self, img, fn_idx, brightness_factor, contrast_factor, saturation_factor, hue_factor):
42
+ for fn_id in fn_idx:
43
+ if fn_id == 0 and brightness_factor is not None:
44
+ img = F.adjust_brightness(img, brightness_factor)
45
+ elif fn_id == 1 and contrast_factor is not None:
46
+ img = F.adjust_contrast(img, contrast_factor)
47
+ elif fn_id == 2 and saturation_factor is not None:
48
+ img = F.adjust_saturation(img, saturation_factor)
49
+ elif fn_id == 3 and hue_factor is not None:
50
+ img = F.adjust_hue(img, hue_factor)
51
+ return img
52
+
53
+ def forward(self, img1, img2):
54
+
55
+ fn_idx, brightness_factor, contrast_factor, saturation_factor, hue_factor = self.get_params(
56
+ self.brightness, self.contrast, self.saturation, self.hue
57
+ )
58
+ img1 = self.jitter_one(img1, fn_idx, brightness_factor, contrast_factor, saturation_factor, hue_factor)
59
+ if torch.rand(1) < self.assymetric_prob: # assymetric:
60
+ fn_idx, brightness_factor, contrast_factor, saturation_factor, hue_factor = self.get_params(
61
+ self.brightness, self.contrast, self.saturation, self.hue
62
+ )
63
+ img2 = self.jitter_one(img2, fn_idx, brightness_factor, contrast_factor, saturation_factor, hue_factor)
64
+ return img1, img2
65
+
66
+ def get_pair_transforms(transform_str, totensor=True, normalize=True):
67
+ # transform_str is eg crop224+color
68
+ trfs = []
69
+ for s in transform_str.split('+'):
70
+ if s.startswith('crop'):
71
+ size = int(s[len('crop'):])
72
+ trfs.append(RandomCropPair(size))
73
+ elif s=='acolor':
74
+ trfs.append(ColorJitterPair(assymetric_prob=1.0, brightness=(0.6, 1.4), contrast=(0.6, 1.4), saturation=(0.6, 1.4), hue=0.0))
75
+ elif s=='': # if transform_str was ""
76
+ pass
77
+ else:
78
+ raise NotImplementedError('Unknown augmentation: '+s)
79
+
80
+ if totensor:
81
+ trfs.append( ToTensorBoth() )
82
+ if normalize:
83
+ trfs.append( NormalizeBoth(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225]) )
84
+
85
+ if len(trfs)==0:
86
+ return None
87
+ elif len(trfs)==1:
88
+ return trfs
89
+ else:
90
+ return ComposePair(trfs)
91
+
92
+
93
+
94
+
95
+
demo/dust3r/croco/demo.py ADDED
@@ -0,0 +1,55 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Copyright (C) 2022-present Naver Corporation. All rights reserved.
2
+ # Licensed under CC BY-NC-SA 4.0 (non-commercial use only).
3
+
4
+ import torch
5
+ from models.croco import CroCoNet
6
+ from PIL import Image
7
+ import torchvision.transforms
8
+ from torchvision.transforms import ToTensor, Normalize, Compose
9
+
10
+ def main():
11
+ device = torch.device('cuda:0' if torch.cuda.is_available() and torch.cuda.device_count()>0 else 'cpu')
12
+
13
+ # load 224x224 images and transform them to tensor
14
+ imagenet_mean = [0.485, 0.456, 0.406]
15
+ imagenet_mean_tensor = torch.tensor(imagenet_mean).view(1,3,1,1).to(device, non_blocking=True)
16
+ imagenet_std = [0.229, 0.224, 0.225]
17
+ imagenet_std_tensor = torch.tensor(imagenet_std).view(1,3,1,1).to(device, non_blocking=True)
18
+ trfs = Compose([ToTensor(), Normalize(mean=imagenet_mean, std=imagenet_std)])
19
+ image1 = trfs(Image.open('assets/Chateau1.png').convert('RGB')).to(device, non_blocking=True).unsqueeze(0)
20
+ image2 = trfs(Image.open('assets/Chateau2.png').convert('RGB')).to(device, non_blocking=True).unsqueeze(0)
21
+
22
+ # load model
23
+ ckpt = torch.load('pretrained_models/CroCo_V2_ViTLarge_BaseDecoder.pth', 'cpu')
24
+ model = CroCoNet( **ckpt.get('croco_kwargs',{})).to(device)
25
+ model.eval()
26
+ msg = model.load_state_dict(ckpt['model'], strict=True)
27
+
28
+ # forward
29
+ with torch.inference_mode():
30
+ out, mask, target = model(image1, image2)
31
+
32
+ # the output is normalized, thus use the mean/std of the actual image to go back to RGB space
33
+ patchified = model.patchify(image1)
34
+ mean = patchified.mean(dim=-1, keepdim=True)
35
+ var = patchified.var(dim=-1, keepdim=True)
36
+ decoded_image = model.unpatchify(out * (var + 1.e-6)**.5 + mean)
37
+ # undo imagenet normalization, prepare masked image
38
+ decoded_image = decoded_image * imagenet_std_tensor + imagenet_mean_tensor
39
+ input_image = image1 * imagenet_std_tensor + imagenet_mean_tensor
40
+ ref_image = image2 * imagenet_std_tensor + imagenet_mean_tensor
41
+ image_masks = model.unpatchify(model.patchify(torch.ones_like(ref_image)) * mask[:,:,None])
42
+ masked_input_image = ((1 - image_masks) * input_image)
43
+
44
+ # make visualization
45
+ visualization = torch.cat((ref_image, masked_input_image, decoded_image, input_image), dim=3) # 4*(B, 3, H, W) -> B, 3, H, W*4
46
+ B, C, H, W = visualization.shape
47
+ visualization = visualization.permute(1, 0, 2, 3).reshape(C, B*H, W)
48
+ visualization = torchvision.transforms.functional.to_pil_image(torch.clamp(visualization, 0, 1))
49
+ fname = "demo_output.png"
50
+ visualization.save(fname)
51
+ print('Visualization save in '+fname)
52
+
53
+
54
+ if __name__=="__main__":
55
+ main()