CapGeo-Bench: Decoupling Visual Perception from Reasoning and Evaluating Geometric Understanding
CapGeo-Bench is proposed, a benchmark of 4,641 high-quality figure-caption pairs equipped with a fine-grained keypoint-based evaluation metric that provides high-quality captions consistently and substantially boosts performance of MLLMs, empirically validating the visual perception bottleneck in geometric reasoning.
Yu-Ying Li, Si-Yi Qian, Hao Liang et al.
· 5 citations