Skip to content

Source paper

MinerU2.5: A Decoupled Vision-Language Model for Efficient High-Resolution Document Parsing

Junbo Niu, Zheng Liu, Zhuangcheng Gu and 58 more

Junbo Niu, Zheng Liu, Zhuangcheng Gu, Bin Wang, Linke Ouyang, Zhiyuan Zhao, Tao Chu, Tianyao He, Fan Wu, Qintong Zhang, Zhenjiang Jin, Guang Liang, Rui Zhang, Wenzheng Zhang, Yuan Qu, Zhifei Ren, Yuefeng Sun, Yuanhong Zheng, Dongsheng Ma, Zirui Tang, Boyu Niu, Ziyang Miao, Hejun Dong, Siyi Qian, Junyuan Zhang, Jingzhou Chen, Fangdong Wang, Xiaomeng Zhao, Liqun Wei, Wei Li, Shasha Wang, Ruiliang Xu, Yuanyuan Cao, Lu Chen, Qianqian Wu, Huaiyu Gu, Lindong Lu, Keming Wang, Dechen Lin, Guanlin Shen, Xuanhe Zhou, Linfeng Zhang, Yuhang Zang, Xiaoyi Dong, Jiaqi Wang, Bo Zhang, Lei Bai, Pei Chu, Weijia Li, Jiang Wu, Lijun Wu, Zhenxiang Li, Guangyu Wang, Zhongying Tu, Chao Xu, Kai Chen, Yu Qiao, Bowen Zhou, Dahua Lin, Wentao Zhang, Conghui He

arXiv 2509.22186 · 2025139 upvotes on Hugging Face

Research ideas from this paper (3)

Generated by AI. The paper’s authors did not write these.

AI summary

MinerU2.5 is a 1.2B-parameter vision-language model designed for efficient, high-resolution document parsing. It achieves state-of-the-art accuracy by decoupling the process into two stages: global layout analysis on a downsampled image and fine-grained content recognition on native-resolution crops, supported by a sophisticated data engine that mines 'hard cases' via inference consistency.

How it works (AI summary)

The model utilizes a NaViT vision encoder and Qwen2-Instruct language decoder within a two-stage pipeline. Stage 1 performs layout detection on a 1036x1036 thumbnail to identify regions; Stage 2 crops these regions from the original high-resolution image and processes them independently for text, table, and formula recognition using specific prompts.