IllusionVQA: A Challenging Optical Illusion Dataset for Vision Language Models

arXiv:2403.15952v1 Announce Type: new
Abstract: The advent of Vision Language Models (VLM) has allowed researchers to investigate the visual understanding of a neural network using natural language. Beyond object classification and detection, VLMs are capable of visual comprehension and common-sense reasoning. This naturally led to the question: How do VLMs respond when the image itself is inherently unreasonable? To this end, we present IllusionVQA: a diverse dataset of challenging optical illusions and hard-to-interpret scenes to test the capability of VLMs …

abstract arxiv beyond classification cs.cl cs.cv dataset detection language language models natural natural language network neural network object optical question reasoning researchers sense type understanding vision visual vlm vlms