Zuming Huang

About Me

Zuming Huang is currently a staff algorithm engineer on the multimodal team at INF Tech since December 2024. Previously, he worked on the multi-modality cognition team at Ant Group starting from April 2019. Before that, he served as an algorithm engineer in the Computer Vision Technology Department at Baidu Inc. He received his Bachelor's degree in Information Engineering from South China University of Technology (SCUT) and his Master's degree in Pattern Recognition and Intelligent Systems from the Institute of Automation, Chinese Academy of Sciences (CASIA). His research interests include Document Intelligence, Large Multimodal Models, and Reinforcement Learning.

Experience & Education

• 2024.12 - Present INF Tech, Multimodal Team, Staff Algorithm Engineer.
• 2019.04 - 2024.12 Ant Group (Hangzhou), Multi-modality Cognition Team, Senior Algorithm Engineer.
• 2017.06 - 2019.04 Baidu Inc. (Beijing), Department of Computer Vision Technology (VIS), Algorithm Engineer.
• 2014.09 - 2017.06 Institute of Automation, Chinese Academy of Sciences (CASIA), M.S. in Pattern Recognition and Intelligence System.
• 2010.09 - 2014.07 South China University of Technology (SCUT), B.Eng. in Information Engineering.

Selected Publications

• Infinity Parser: Layout Aware Reinforcement Learning for Scanned Document Parsing
  Baode Wang, Biao Wu, Weizhen Li, Meng Fang, Zuming Huang, Jun Huang, Haozhe Wang, Yanjie Liang, Ling Chen, Wei Chu, and Yuan Qi
  Technical Report, 2025. (Project Lead)
  Paper / Code / Dataset / Model
• VL-Rethinker: Incentivizing Self-Reflection of Vision-Language Models with Reinforcement Learning
  Haozhe Wang, Chao Qu, Zuming Huang, Wei Chu, Fangzhen Lin, and Wenhu Chen
  Conference on Neural Information Processing Systems (NeurIPS), 2025. (CCF-A, Spotlight)
  Paper / Code / Dataset / Model
• Fine-grained Pesudo Labels for Scene Text Recognition
  Xiaoyu Li, Xiaoxue Chen, Zuming Huang, Lele Xie, Jingdong Chen, and Ming Yang
  ACM International Conference on Multimedia (ACM MM), 2023. (CCF-A)
  Paper
• Look More Than Once: An Accurate Detector for Text of Arbitrary Shapes
  Chengquan Zhang*, Borong Liang*, Zuming Huang*, Mengyi En, Junyu Han, Errui Ding, and Xinghao Ding
  Computer Vision and Pattern Recognition Conference (CVPR), 2019. (CCF-A, Equal first authors)
  Paper
• A Single-shot Arbitrarily-shaped Text Detector based on Context Attended Multi-task Learning
  Pengfei Wang, Chengquan Zhang, Fei Qi, Zuming Huang, Mengyi En, Junyu Han, Jingtuo Liu, Errui Ding, Guangming Shi
  ACM International Conference on Multimedia (ACM MM), 2019. (CCF-A, Integrated into PaddleOCR)
  Paper / Code
• TextNet: Irregular Text Reading from Images with an End-to-End Trainable Network
  Yipeng Sun, Chengquan Zhang, Zuming Huang, Jiaming Liu, Junyu Han, and Errui Ding
  Asian Conference on Computer Vision (ACCV), 2018. (Oral)
  Paper
• Building Extraction from Multi-source Remote Sensing Images via Deep Deconvolution Neural Networks
  Zuming Huang, Guangliang Cheng, Hongzhen Wang, Haichang Li, Limin Shi, and Chunhong Pan
  International Geoscience and Remote Sensing Symposium (IGARSS), 2016.
  Paper
• Extraction of Virtual Baselines from Distorted Document Images using Curvilinear Projection
  Gaofeng Meng, Zuming Huang, Yonghong Song, Shiming Xiang, and Chunhong Pan.
  International Conference on Computer Vision (ICCV), 2015. (CCF-A)
  Paper

Honors & Awards

• Employee of the Year, INF Tech, 2025
• CCF Science and Technology Award, 2023
• Winning Team, ICDAR Competition on SVRD, 2023
• Honorable Mentor, Ant Group, 2021
• Excellent Project, Baidu Vis, 2018
• Best New Employee Award, Baidu Vis, 2017
• Outstanding Student Cadre Award, CASIA, 2016
• National Scholarship, CASIA, 2016
• Best Paper Award, ICCC, 2014
• Student Award of Merit, SCUT, 2013
• National Scholarship, SCUT, 2012

Personal Interest

I'm a massive football fan. I've been playing and following the game since middle school.