Metraipolis › archive › dc_4c5efa2446cb
Why VLMs Miss Small Objects, and When Zooming In Is Safe — arXiv:2610.09313
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2610.09313
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-08T12:47:05.338Z
- SHA-256 of the archived text
e5dd1720977d85db564539a919465e23477ba21f0fbaab3ac41b0658816466f8- Stable URL of this page
https://metraipolis.com/documento/dc_4c5efa2446cb
How to cite
Why VLMs Miss Small Objects, and When Zooming In Is Safe — arXiv:2610.09313 — arXiv (en), https://arxiv.org/abs/2610.09313 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_4c5efa2446cb (Metraipolis archive dc_4c5efa2446cb, sha256 e5dd1720977d85db564539a919465e23477ba21f0fbaab3ac41b0658816466f8)
BibTeX
@misc{dc4c5efa2446cb,
title = {Why VLMs Miss Small Objects, and When Zooming In Is Safe — arXiv:2610.09313},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2610.09313},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_4c5efa2446cb; sha256 e5dd1720977d85db564539a919465e23477ba21f0fbaab3ac41b0658816466f8},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.09313. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Vision-language models (VLMs) often miss small objects in large images. We ask three questions: what limits them, which of these limits better models can remove, and whether the classical way of handling large images, local decomposition, still has a future. We answer them with a theory built on two quantities of the image interface: S, the number of visual tokens across an object's side, and L, the content a call must cover. The limits: a W x H image sent whole within N tokens gives an object of side m at most m*sqrt(N/(WH)) tokens per side. Doubling the token budget N therefore raises the largest S a whole image can reach by only 41%, and seeing an image at the S an object needs costs at least order S^2 tokens whatever the model. If recognition improves gradually with S, any search strategy, zoom agents included, obeys a recall-cost frontier. What better models can change: the S an object needs and how much one call can carry, measured in bits per object found; the coverage cost remains. Decomposition: yes. Assuming only that more tokens per object and less content per call do not hurt on average, splitting an image cannot lower recall if no view zooms out relative to the whole image and views overlap by one object. Neither part of this condition can be dropped; we bound the cost of every such decomposition, and a simple rule approaches the bound as the image grows. We test the theory in about 177,000 requests on 797 images. On controlled images, none of 40 orderings predicted for 8 VLMs is violated. Checked after the fact on drawings, floor plans, natural and synthetic images, 61 of 89 implied orderings hold significantly and 4 fail, all for OpenAI models given more pixels than their default path. On construction drawings the rule never significantly lowered recall relative to the whole image and raised it by up to 0.28. Code and data: this https URL Tema degli esploratori: statistica e scienza dei dati (elenco arXiv cs.CV, nuovi lavori del giorno). Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV). Pagina del lavoro: https://arxiv.org/abs/2610.09313 — 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 2026-10-08T12:18:15.327Z dalla pagina https://arxiv.org/list/cs.CV/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).