CORTEXA
← Browse

Chengyang Zhang

2 papers indexed

arxivcs.CV2026-07-23

Do Pathology Vision-Language Models Truly See Pathology?

Chengyang Zhang, Wenchuan Zhang, Bo Li, Xinyu Liu, Jiaming Yang, Mengran Li, et al.

Pathology vision-language models (VLMs) have recently progressed rapidly and are commonly evaluated by answer accuracy on pathology VQA benchmarks. However, we dig into current evaluations and identify three overlooked issues: 1) Visual evidence is not always necessary. For insta…

View free PDFSource page
arxivcs.CV2026-07-01

HieDG: A Hierarchical Discrete Geometry-Guided Framework for Multi-Animal Tracking

Chenxun Deng, Zhongde Zhang, Ye Yuan, Chengyang Zhang, Yifan Zhang, Bohao Chen, et al.

Multi-animal tracking (MAT) is critical for wildlife monitoring and behavioral analysis, yet remains challenging due to uniform appearance, high density, and irregular motion. Existing methods typically follow heuristic- or query-based paradigms: the former relies on handcrafted…

View free PDFSource page