Hidden in Plain Sight: Evaluating Abstract Shape Recognition in Vision-Language Models
This work addresses limitations in visual perception for AI systems, particularly in shape recognition, which is incremental as it builds on prior research on VLM reliance on shape.
The paper tackles the problem of shape recognition in Vision-Language Models (VLMs) by introducing IllusionBench, a dataset that challenges VLMs to decipher shapes from visual element arrangements, revealing that current VLMs struggle with this task while humans detect them easily.
Despite the importance of shape perception in human vision, early neural image classifiers relied less on shape information for object recognition than other (often spurious) features. While recent research suggests that current large Vision-Language Models (VLMs) exhibit more reliance on shape, we find them to still be seriously limited in this regard. To quantify such limitations, we introduce IllusionBench, a dataset that challenges current cutting-edge VLMs to decipher shape information when the shape is represented by an arrangement of visual elements in a scene. Our extensive evaluations reveal that, while these shapes are easily detectable by human annotators, current VLMs struggle to recognize them, indicating important avenues for future work in developing more robust visual perception systems. The full dataset and codebase are available at: \url{https://arshiahemmat.github.io/illusionbench/}