作者位次 #9/20ACL 2026Poster· Natural Language Processing
Decoding Scientific Experimental Images: The SPUR Benchmark for Perception, Understanding, and Reasoning
Junpeng Ding · Zichen Tang · Haihong E · Mengyuan Ji · Yang Liu · Haolin Tian · Haiyang Sun · Pengqi Sun · Yang Xu · Yichen Liu · Haocheng Gao · Zijie Xi · Ruomeng Jiang · Peizhi Zhao · Rongjin Li · Yuanze Li · Jiacheng Liu · Zhongjun Yang · Jintong Chen · Siying Lin
摘要
We introduce SPUR, a comprehensive benchmark for scientific experimental image perception, understanding, and reasoning, comprising 4,264 question-answering (QA) pairs derived from 1,084 expert-curated images. SPUR features three key innovations: (1) Panel-Level Fine-Grained Perception: evaluatin...