Vision-language models classify proxemic danger in egocentric robot images only slightly better than random, except Qwen-VL which detects high-danger scenes with high recall that is not tied to accurate person localization.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Can Vision-Language Models Assess Proxemic Risk from Egocentric Robot Images?
Vision-language models classify proxemic danger in egocentric robot images only slightly better than random, except Qwen-VL which detects high-danger scenes with high recall that is not tied to accurate person localization.