Abstract:Infants learn not only object categories but also fine-grained visual attributes such as color, size, and texture from limited experience. Prior infant-scale vision--language models have mainly been evaluated on object recognition, leaving open whether they support within-class attribute discrimination. We introduce a controlled benchmark that varies color, size, and texture across 67 everyday object classes using synthetic rendering to decouple attribute values from object identity. We evaluate infant-trained models (CVCL and an infant-trained DINO baseline) against web-scale and ImageNet models (CLIP, SigLIP, ResNeXt) under two complementary settings: an image-only prototype test and a text--vision test with attribute--object prompts. We find a dissociation between visual and linguistic attribute information: infant-trained models form strong visual representations for size and discriminate texture comparably to other models, but perform poorly on visual color discrimination, and in the text--vision setting they struggle to ground color and show only modest size grounding. In contrast, web-trained vision--language models strongly ground color from text while exhibiting weaker visual size discrimination.
| Subjects: | Machine Learning (cs.LG) |
| Cite as: | arXiv:2512.18951 [cs.LG] |
| (or arXiv:2512.18951v3 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2512.18951 arXiv-issued DOI via DataCite |
From: Patrick Batsell [view email]
[v1]
Mon, 22 Dec 2025 01:58:17 UTC (2,441 KB)
[v2]
Thu, 26 Mar 2026 01:12:36 UTC (898 KB)
[v3]
Wed, 13 May 2026 02:13:34 UTC (899 KB)
此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。 原文来自 — 版权归原作者所有。