RyanWW commited on
Commit
80904ea
·
1 Parent(s): 81fb5c0

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +78 -1
README.md CHANGED
@@ -11,7 +11,84 @@ pretty_name: Spatial457
11
  size_categories:
12
  - 10K<n<100K
13
  ---
 
 
 
 
 
 
 
 
14
 
15
- # Spatial457: A Diagnostic Benchmark for 6D Spatial Reasoning of Large Multimodal Models
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
16
 
17
 
 
11
  size_categories:
12
  - 10K<n<100K
13
  ---
14
+ e
15
+ <!-- # Spatial457: A Diagnostic Benchmark for 6D Spatial Reasoning of Large Multimodal Models -->
16
+ e
17
+ <h1 align="center">
18
+ <a href="https://arxiv.org/abs/2502.08636">
19
+ Spatial457: A Diagnostic Benchmark for 6D Spatial Reasoning of Large Multimodal Models
20
+ </a>
21
+ </h1>
22
 
23
+ <p align="center">
24
+ <a href=".">Xingrui Wang</a><sup>1</sup>,
25
+ <a href=".">Wufei Ma</a><sup>1</sup>,
26
+ <a href=".">Tiezheng Zhang</a><sup>1</sup>,
27
+ <a href=".">Celso M de Melo</a><sup>2</sup>,
28
+ <a href=".">Jieneng Chen</a><sup>1</sup>,
29
+ <a href=".">Alan Yuille</a><sup>1</sup>
30
+ </p>
31
+
32
+ <p align="center">
33
+ <sup>1</sup> Johns Hopkins University &nbsp;&nbsp;
34
+ <sup>2</sup> DEVCOM Army Research Laboratory
35
+ </p>
36
+
37
+
38
+ <p align="center">
39
+ <a href="https://xingruiwang.github.io/projects/Spatial457/">Project Page</a> /
40
+ <a href="https://arxiv.org/abs/2502.08636">Paper</a> /
41
+ <a href="https://huggingface.co/datasets/RyanWW/Spatial457">🤗 Huggingface</a> /
42
+ <a href="https://github.com/XingruiWang/Spatial457">Code</a>
43
+ </p>
44
+
45
+
46
+ <p align="center">
47
+ <img src="https://github.com/XingruiWang/Spatial45r/imgs/teaser.png?raw=True" alt="Spatial457 Teaser" width="80%">
48
+ </p>
49
+
50
+ <!-- <p align="center"><i>
51
+ Official implementation of the CVPR 2025 (Highlight) paper:
52
+ <strong>Spatial457: A Diagnostic Benchmark for 6D Spatial Reasoning of Large Multimodal Models</strong>
53
+ </i></p> -->
54
+
55
+ ---
56
+
57
+ ## 🧠 Introduction
58
+
59
+ Spatial457 is a diagnostic benchmark designed to evaluate the 6D spatial reasoning capabilities of large multimodal models (LMMs). It systematically introduces four key capabilities—multi-object understanding, 2D and 3D localization, and 3D orientation—across five difficulty levels and seven question types, progressing from basic recognition to complex physical interaction.
60
+
61
+ ---
62
+
63
+ ## 📦 Download
64
+
65
+ You can access the full dataset and evaluation toolkit:
66
+
67
+ - **Dataset**: [Hugging Face](https://huggingface.co/datasets/RyanWW/Spatial457)
68
+ - **Code**: [GitHub Repository](https://github.com/XingruiWang/Spatial457)
69
+ - **Paper**: [arXiv 2502.08636](https://arxiv.org/abs/2502.08636)
70
+
71
+ ---
72
+
73
+ ## 📊 Benchmark
74
+
75
+ We benchmarked a wide range of state-of-the-art models—including GPT-4o, Gemini, Claude, and several open-source LMMs—on all subsets. Performance consistently drops as task difficulty increases. PO3D-VQA and humans remain most robust across all levels.
76
+
77
+ The table below summarizes model performance across 7 subsets:
78
+
79
+ <!-- Include table image or markdown table here if needed -->
80
+
81
+ ---
82
+
83
+ ## 📚 Citation
84
+
85
+ ```bibtex
86
+ @inproceedings{wang2025spatial457,
87
+ title = {Spatial457: A Diagnostic Benchmark for 6D Spatial Reasoning of Large Multimodal Models},
88
+ author = {Wang, Xingrui and Ma, Wufei and Zhang, Tiezheng and de Melo, Celso M and Chen, Jieneng and Yuille, Alan},
89
+ booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
90
+ year = {2025},
91
+ url = {https://arxiv.org/abs/2502.08636}
92
+ }
93
 
94