From Pixel to Coding: Evaluating the Figure Reproduction Capabilities of MLLMs

作者：Zijian Chen, Zhengyu Chen, Bohan Liang, Lirong Deng, Yushuo Zheng, Yanwei Jiang, Qi Jia, Kaiwei Zhang, Wenjun Zhang, Guangtao Zhai · 单位：Institute of Image Communication and Information Processing, Shanghai Jiao Tong University, Shanghai, 200240, China, Shanghai Artificial Intelligence Laboratory, Shanghai, 200030, China, Macao Polytechnic University, Macao, 999078, China · 会议/期刊：arXiv preprint · 方向：cs.CV · 发布日期：2026-10-08