image-degradation

Tag

Cards List
#image-degradation

Small Vision-Language Models Know When They Are Wrong But Cannot Say So: A Two-Model Study of Stated versus Internal Confidence Under Realistic Image Degradation

arXiv cs.CL ↗ · 2026-07-27 Cached

This paper evaluates how small open-weight vision-language models (Qwen2-VL-2B and SmolVLM) handle realistic image degradations, finding that their verbalized confidence is unreliable while internal token probability provides much better error detection, though both fail under severe low-light conditions.

0 favorites 0 likes
← Back to home

Submit Feedback