GlyphPattern: An Abstract Pattern Recognition Benchmark for Vision-Language Models

TL;DR

GlyphPattern is a new benchmark dataset designed to evaluate the ability of vision-language models to recognize and understand abstract visual patterns described in natural language, revealing significant challenges for current models.

Contribution

We introduce GlyphPattern, a comprehensive dataset and evaluation framework for abstract pattern recognition in vision-language models, highlighting their limitations in understanding complex visual patterns.

Findings

Current VLMs perform poorly on GlyphPattern with only 55% accuracy.

Few-shot prompting yields only marginal improvements.

Challenges exist in visual processing, language understanding, and pattern generalization.

Abstract

Vision-Language Models (VLMs) building upon the foundation of powerful large language models have made rapid progress in reasoning across visual and textual data. While VLMs perform well on vision tasks that they are trained on, our results highlight key challenges in abstract pattern recognition. We present GlyphPattern, a 954 item dataset that pairs 318 human-written descriptions of visual patterns from 40 writing systems with three visual presentation styles. GlyphPattern evaluates abstract pattern recognition in VLMs, requiring models to understand and judge natural language descriptions of visual patterns. GlyphPattern patterns are drawn from a large-scale cognitive science investigation of human writing systems; as a result, they are rich in spatial reference and compositionality. Our experiments show that GlyphPattern is challenging for state-of-the-art VLMs (GPT-4o achieves only 55% accuracy), with marginal gains from few-shot prompting. Our detailed error analysis reveals challenges at multiple levels, including visual processing, natural language understanding, and pattern generalization.