Hugging Face's logo

lc700x
/
patchfusion-depth-anything-small-hf

Safetensors
depth_anything
Model card Files Community
lc700x commited on
Commit
745be29
·
verified ·
1 Parent(s): 2671e32

Upload 2 files

Browse files
Files changed (2) hide show
  1. config.json +86 -171
  2. preprocessor_config.json +27 -0
config.json CHANGED
@@ -1,171 +1,86 @@
1
- {
2
- "coarse_branch": {
3
- "attractor_alpha": 1000,
4
- "attractor_gamma": 2,
5
- "attractor_kind": "mean",
6
- "attractor_type": "inv",
7
- "aug": true,
8
- "bin_centers_type": "softplus",
9
- "bin_embedding_dim": 128,
10
- "clip_grad": 0.1,
11
- "dataset": "nyu",
12
- "depth_anything": true,
13
- "distributed": true,
14
- "do_resize": false,
15
- "force_keep_ar": true,
16
- "freeze_midas_bn": true,
17
- "gpu": "NULL",
18
- "img_size": [
19
- 392,
20
- 518
21
- ],
22
- "inverse_midas": false,
23
- "log_images_every": 0.1,
24
- "max_depth": 80,
25
- "max_temp": 50.0,
26
- "max_translation": 100,
27
- "memory_efficient": true,
28
- "midas_model_type": "vits",
29
- "min_depth": 0.001,
30
- "min_temp": 0.0212,
31
- "model": "zoedepth",
32
- "n_attractors": [
33
- 16,
34
- 8,
35
- 4,
36
- 1
37
- ],
38
- "n_bins": 64,
39
- "name": "ZoeDepth",
40
- "notes": "",
41
- "output_distribution": "logbinomial",
42
- "prefetch": false,
43
- "pretrained_resource": "local::./work_dir/DepthAnything_vits.pt",
44
- "print_losses": false,
45
- "project": "ZoeDepth",
46
- "random_crop": false,
47
- "random_translate": false,
48
- "root": ".",
49
- "save_dir": "",
50
- "shared_dict": "NULL",
51
- "tags": "",
52
- "train_midas": true,
53
- "translate_prob": 0.2,
54
- "type": "DA-ZoeDepth",
55
- "uid": "NULL",
56
- "use_amp": false,
57
- "use_pretrained_midas": true,
58
- "use_shared_dict": false,
59
- "validate_every": 0.25,
60
- "version_name": "v1",
61
- "workers": 16
62
- },
63
- "fine_branch": {
64
- "attractor_alpha": 1000,
65
- "attractor_gamma": 2,
66
- "attractor_kind": "mean",
67
- "attractor_type": "inv",
68
- "aug": true,
69
- "bin_centers_type": "softplus",
70
- "bin_embedding_dim": 128,
71
- "clip_grad": 0.1,
72
- "dataset": "nyu",
73
- "depth_anything": true,
74
- "distributed": true,
75
- "do_resize": false,
76
- "force_keep_ar": true,
77
- "freeze_midas_bn": true,
78
- "gpu": "NULL",
79
- "img_size": [
80
- 392,
81
- 518
82
- ],
83
- "inverse_midas": false,
84
- "log_images_every": 0.1,
85
- "max_depth": 80,
86
- "max_temp": 50.0,
87
- "max_translation": 100,
88
- "memory_efficient": true,
89
- "midas_model_type": "vits",
90
- "min_depth": 0.001,
91
- "min_temp": 0.0212,
92
- "model": "zoedepth",
93
- "n_attractors": [
94
- 16,
95
- 8,
96
- 4,
97
- 1
98
- ],
99
- "n_bins": 64,
100
- "name": "ZoeDepth",
101
- "notes": "",
102
- "output_distribution": "logbinomial",
103
- "prefetch": false,
104
- "pretrained_resource": "local::./work_dir/DepthAnything_vits.pt",
105
- "print_losses": false,
106
- "project": "ZoeDepth",
107
- "random_crop": false,
108
- "random_translate": false,
109
- "root": ".",
110
- "save_dir": "",
111
- "shared_dict": "NULL",
112
- "tags": "",
113
- "train_midas": true,
114
- "translate_prob": 0.2,
115
- "type": "DA-ZoeDepth",
116
- "uid": "NULL",
117
- "use_amp": false,
118
- "use_pretrained_midas": true,
119
- "use_shared_dict": false,
120
- "validate_every": 0.25,
121
- "version_name": "v1",
122
- "workers": 16
123
- },
124
- "guided_fusion": {
125
- "g2l": true,
126
- "in_channels": [
127
- 32,
128
- 64,
129
- 64,
130
- 64,
131
- 64,
132
- 64
133
- ],
134
- "n_channels": 5,
135
- "num_patches": [
136
- 203056,
137
- 66304,
138
- 16576,
139
- 4144,
140
- 1036,
141
- 266
142
- ],
143
- "patch_process_shape": [
144
- 392,
145
- 518
146
- ],
147
- "type": "GuidedFusionPatchFusion"
148
- },
149
- "image_raw_shape": [
150
- 2160,
151
- 3840
152
- ],
153
- "max_depth": 80,
154
- "min_depth": 0.001,
155
- "patch_process_shape": [
156
- 392,
157
- 518
158
- ],
159
- "patch_split_num": [
160
- 4,
161
- 4
162
- ],
163
- "pretrain_model": [
164
- "./work_dir/depthanything_vits_u4k/coarse_pretrain/checkpoint_24.pth",
165
- "./work_dir/depthanything_vits_u4k/fine_pretrain/checkpoint_24.pth"
166
- ],
167
- "sigloss": {
168
- "type": "SILogLoss"
169
- },
170
- "transformers_version": "4.36.2"
171
- }
 
1
+ {
2
+ "architectures": [
3
+ "DepthAnythingForDepthEstimation"
4
+ ],
5
+ "backbone": null,
6
+ "backbone_config": {
7
+ "apply_layernorm": true,
8
+ "architectures": [
9
+ "Dinov2Model"
10
+ ],
11
+ "attention_probs_dropout_prob": 0.0,
12
+ "drop_path_rate": 0.0,
13
+ "hidden_act": "gelu",
14
+ "hidden_dropout_prob": 0.0,
15
+ "hidden_size": 384,
16
+ "image_size": 518,
17
+ "initializer_range": 0.02,
18
+ "layer_norm_eps": 1e-06,
19
+ "layerscale_value": 1.0,
20
+ "mlp_ratio": 4,
21
+ "model_type": "dinov2",
22
+ "num_attention_heads": 6,
23
+ "num_channels": 3,
24
+ "num_hidden_layers": 12,
25
+ "out_features": [
26
+ "stage9",
27
+ "stage10",
28
+ "stage11",
29
+ "stage12"
30
+ ],
31
+ "out_indices": [
32
+ 9,
33
+ 10,
34
+ 11,
35
+ 12
36
+ ],
37
+ "patch_size": 14,
38
+ "qkv_bias": true,
39
+ "reshape_hidden_states": false,
40
+ "stage_names": [
41
+ "stem",
42
+ "stage1",
43
+ "stage2",
44
+ "stage3",
45
+ "stage4",
46
+ "stage5",
47
+ "stage6",
48
+ "stage7",
49
+ "stage8",
50
+ "stage9",
51
+ "stage10",
52
+ "stage11",
53
+ "stage12"
54
+ ],
55
+ "torch_dtype": "float32",
56
+ "use_mask_token": true,
57
+ "use_swiglu_ffn": false
58
+ },
59
+ "backbone_kwargs": null,
60
+ "depth_estimation_type": "relative",
61
+ "fusion_hidden_size": 64,
62
+ "head_hidden_size": 32,
63
+ "head_in_index": -1,
64
+ "initializer_range": 0.02,
65
+ "max_depth": 1,
66
+ "model_type": "depth_anything",
67
+ "neck_hidden_sizes": [
68
+ 48,
69
+ 96,
70
+ 192,
71
+ 384
72
+ ],
73
+ "output_attentions": false,
74
+ "patch_size": 14,
75
+ "reassemble_factors": [
76
+ 4,
77
+ 2,
78
+ 1,
79
+ 0.5
80
+ ],
81
+ "reassemble_hidden_size": 384,
82
+ "torch_dtype": "float32",
83
+ "transformers_version": null,
84
+ "use_pretrained_backbone": false,
85
+ "use_timm_backbone": false
86
+ }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
preprocessor_config.json ADDED
@@ -0,0 +1,27 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "do_normalize": true,
3
+ "do_pad": false,
4
+ "do_reduce_labels": false,
5
+ "do_rescale": true,
6
+ "do_resize": true,
7
+ "ensure_multiple_of": 14,
8
+ "image_mean": [
9
+ 0.485,
10
+ 0.456,
11
+ 0.406
12
+ ],
13
+ "image_processor_type": "DPTImageProcessor",
14
+ "image_std": [
15
+ 0.229,
16
+ 0.224,
17
+ 0.225
18
+ ],
19
+ "keep_aspect_ratio": true,
20
+ "resample": 3,
21
+ "rescale_factor": 0.00392156862745098,
22
+ "size": {
23
+ "height": 518,
24
+ "width": 518
25
+ },
26
+ "size_divisor": null
27
+ }