同一空間か否か：視点間矛盾による空間知能のベンチマーク
Same Space or Not: Benchmarking Spatial Intelligence Through Cross-View Contradictions
3視点の同一シーンと1視点の編集済みシーンを提示し、矛盾する視点を特定・説明させる空間知能ベンチマークSSoNを提案。最強VLMでも人間の83.3%に対し39.7%と大きく劣ることを示した。
詳しい要約
1. どんなもの？
2. 先行研究と比べてどこがすごい？
3. 技術・手法の肝は？
4. どうやって有効だと検証した？
5. 議論はある？
6. 次に読むべき論文は？
※ AIが要旨から生成した要約です。正確性は原文をご確認ください。
著者: Huizi Cao, Guanyu Xu, Zhixuan Liang, Ming Yin, Jiuxiang Gu, Manling Li, Mengdi Wang, Shilong Liu
分類: cs.CV, cs.LG
原文アブストラクト
Spatial intelligence, the capacity to represent and reason about physical space, underlies embodied perception and real-world visual understanding. Existing benchmarks probe it through spatial relations, perspective-taking, and mental modeling. What remains underexamined is whether a model can maintain a coherent scene representation across views, since pairwise comparison alone cannot reveal which view departs from the truth. We thus introduce Same Space or Not (SSoN), a benchmark of cross-view contradiction resolution comprising 1,000 questions from seven indoor scene sources and six categories of controlled 3D edits. Each question presents three views of an unchanged scene and one view of an edited variant. Models must identify the inconsistent view and explain the contradiction without a designated trusted reference or a marked target. Our pipeline iteratively translates human review into generation constraints. Automated checks and exhaustive human validation screen out both trivial visual shortcuts and questions lacking sufficient evidence for a unique answer. Humans achieve 83.3% accuracy, compared with 39.7% for the strongest vision-language models (VLMs), falling to 34.9% when a valid visual explanation is also required. We investigate prompting, camera information, and 3D reconstruction as interventions. On a 100-question intervention set with a 24% baseline, prompting and 3D reconstruction remain below 50% answer accuracy, except when ground-truth camera information is supplied, which reaches 61% answer accuracy but only 48% True solve accuracy. On a held-out 201-question test set, Qwen2.5-VL-7B achieves 21.89% accuracy, 33.33% after answer-only fine-tuning, and 41.79% after fine-tuning on both rationales and answers. These results identify cross-view contradiction resolution as a substantial weakness of current VLMs that can be partly addressed by targeted supervision.
