Metraipolis › archive › dc_907ff010b985
From Reasoning Failures to Composable Video Spatial Intelligence — arXiv:2610.01999
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2610.01999
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-06T19:15:21.437Z
- SHA-256 of the archived text
6050cf08ac315edc41240f3859e48866bcaf19993303f8c08abbdf8ad69c730e- Stable URL of this page
https://metraipolis.com/documento/dc_907ff010b985
How to cite
From Reasoning Failures to Composable Video Spatial Intelligence — arXiv:2610.01999 — arXiv (en), https://arxiv.org/abs/2610.01999 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_907ff010b985 (Metraipolis archive dc_907ff010b985, sha256 6050cf08ac315edc41240f3859e48866bcaf19993303f8c08abbdf8ad69c730e)
BibTeX
@misc{dc907ff010b985,
title = {From Reasoning Failures to Composable Video Spatial Intelligence — arXiv:2610.01999},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2610.01999},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_907ff010b985; sha256 6050cf08ac315edc41240f3859e48866bcaf19993303f8c08abbdf8ad69c730e},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.01999. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Spatial reasoning benchmarks evaluate vision-language models across diverse tasks, but task-level scores do not reveal which underlying capabilities account for success or failure. Each task requires recovering spatial evidence, representing geometry, and reasoning over it. We disentangle these capabilities by comparing predicted and ground-truth spatial context under a shared schema and coordinate contract. This comparison reveals four recurring sources of error: inaccurate perception, missing information in the spatial context, selection of the wrong measurement, and errors in reference frames or in tracking position and orientation. Guided by this diagnosis, we develop CROSS, a training-free library of typed geometric operators and spatial skills that function over available evidence to support reliable video spatial reasoning. The resulting library supplies verified context to non-coding VLMs or callable skills to a SpatialClaw agent. We evaluate \methodname{} on five benchmarks. \methodname{} raises the average score from 55.9\% to 60.2\% on ReVSI and improves the SpatialClaw result from 62.8\% to 66.3\% on DSI-Bench. These gains demonstrate that explicit handling of spatial conventions can repair systematic reasoning failures without additional training.
Tema degli esploratori: statistica e scienza dei dati (retro-catalogo arXiv: annunci del 2026-10-02 in cs.CV).
Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV).
Pagina del lavoro: https://arxiv.org/abs/2610.01999 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-06T17:37:36.874Z dalla pagina https://arxiv.org/catchup/cs.CV/2026-10-02?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).