MEBench is a new benchmark and data-generation pipeline that measures mutual exclusivity bias in VLMs, finding weak bias but some use of spatial context to resolve novel-object ambiguity.
Uouo: Uncontextualized uncommon objects for measuring knowledge horizons of vision language models
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2025 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
MEBench: A Novel Benchmark for Understanding Mutual Exclusivity Bias in Vision-Language Models
MEBench is a new benchmark and data-generation pipeline that measures mutual exclusivity bias in VLMs, finding weak bias but some use of spatial context to resolve novel-object ambiguity.