Do MLLMs Exhibit Human-like Perceptual Behaviors? HVSBench: A Benchmark for MLLM Alignment with Human Perceptual Behavior

Jiaying Lin^*, Shuquan Ye^*, Dan Xu, Wanli Ouyang, Rynson W.H. Lau.
IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2026 (Findings).
project arxiv