COMEX: A Composition-Grounded Benchmark and Learning Framework for Explainable Aesthetic Image Cropping

作者：Rui Yang, Wei Zhou, Dingyong Gou, Xiaohui Cui, Cong Li, Yinyin Gong, Yipo Huang, Jiliang Zhao · 单位：State Key Laboratory of Mobile Network and Mobile Multimedia Technology, ZTE Corporation, School of Data Science and Institute of Artificial Intelligence, Chang’an University, State Key Laboratory of Mobile Network and Mobile Multimedia Technology, ZTE Corporation Shenzhen China, School of Data Science and Institute of Artificial Intelligence, Chang’an University Xi’an China · 会议/期刊：arXiv preprint · 方向：cs.CV · 发布日期：2026-08-11