Jiaxin Zhang

I am a researcher at Huawei.

I received my Ph.D., M.S., and B.S. degrees from South China University of Technology (SCUT), where I was advised by Prof. Lianwen Jin.

Previously, I interned at Huawei Cloud, IntSig (CamScanner), and LG AI Research.

eejxzhang@gmail.com / Google Scholar / GitHub

News

2026Q-DRFormer accepted to ACM Transactions on Graphics.
2026Three papers accepted to Pattern Recognition: DocAligner, AutoScaler, and Text Prior-Guided Cell Alignment.

Papers

Selected first-author work.

Vision-Language & Multimodal Models

DocKylin: A Large Multimodal Model for Visual Document Understanding with Efficient Visual Slimming

Jiaxin Zhang^†, Wentao Yang^†, Songxuan Lai, Zecheng Xie, Lianwen Jin*
^† Equal contribution. ^* Corresponding author.

AAAI, 2025

Paper / Code

Smaller but Better: Unifying Layout Generation with Smaller Large Language Models

P. Zhang, Jiaxin Zhang, J. Cao, H. Li, L. Jin

International Journal of Computer Vision, 133(7), 3891–3917, 2025

Paper / Code

Decoder-only vs cross-attention architectures

RedundancyLens: Revealing and Exploiting Visual Token Processing Redundancy for Efficient Decoder-Only MLLMs

H. Li, Jiaxin Zhang, W. Liao, D. Peng, K. Ding, L. Jin

Findings of ACL, 2025

Paper / Code

Image Restoration & Enhancement

DocAligner: Automating the Annotation of Photographed Documents through Real-Virtual Alignment

Jiaxin Zhang, P. Zhang, H. Cheng, X. Chen, H. Xu, K. Ding, L. Jin

Pattern Recognition, 2026

Paper / Code

Q-DRFormer: Query-Position Mapping Flow Prediction with Progressive Refinement for Document Rectification

J. Li, Jiaxin Zhang, J. Guo, Q. Zhang, X. Zheng, Q. Yin, W. Zeng

ACM Transactions on Graphics, 46(2), 1–16, 2026

Paper / Code

Towards Real-World Document Specular Highlight Removal: The DocHighlight Dataset and DocSHRNet Method

H. Xu, Jiaxin Zhang, H. Cheng, P. Zhang, X. Zheng, L. Jin

PRCV, 109–124, 2025

Paper / Code

Enhancing Document Dewarping Evaluation: A New Metric with Improved Accuracy and Efficiency

Jiaxin Zhang, P. Zhang, D. Peng, H. Xu, L. Jin

Pattern Recognition Letters, 195, 51–58, 2025

Paper / Code

OCRGenBench: A Comprehensive Benchmark for Evaluating OCR Generative Capabilities

P. Zhang, H. Xu, Jiaxin Zhang, X. Zheng, G. Xu, Y. Zhang, J. Liu, Z. Yang, W. Zhou, et al.

arXiv 2025

Paper / Code

DocRes: A Generalist Model Toward Unifying Document Image Restoration Tasks

Jiaxin Zhang, Dezhi Peng, Chongyu Liu, Peirong Zhang, Lianwen Jin*

CVPR, 2024

Paper / Code

Appearance Enhancement for Camera-Captured Document Images in the Wild

Jiaxin Zhang, L. Liang, K. Ding, F. Guo, L. Jin

IEEE Transactions on Artificial Intelligence, 5(5), 2319–2330, 2023

Paper / Code

Marior: Margin Removal and Iterative Content Rectification for Document Dewarping in the Wild

Jiaxin Zhang, C. Luo, L. Jin, F. Guo, K. Ding

ACM MM, 2022

Paper / Code

OCR

Text Prior-Guided Cell Alignment with Transformer for Robust Table Structure Recognition

J. Li, Jiaxin Zhang, Q. Yin, Q. Zhang, X. Zheng, W. Zeng, P. Guo

Pattern Recognition, 183, 114832, 2026

Paper / Code

AutoScaler: Self Scale Alignment for Handwritten Mathematical Expression Recognition

W. Yang, Jiaxin Zhang, L. Jin

Pattern Recognition, 2026

Paper / Code

QT-TextSR: Enhancing Scene Text Image Super-Resolution via Efficient Interaction with Text Recognition

C. Liu, Q. Jiang, D. Peng, Y. Kong, Jiaxin Zhang, L. Xiong, J. Duan, C. Sun, L. Jin

Neurocomputing, 620, 129241, 2025

Paper / Code

Towards Accurate Readings of Water Meters by Eliminating Transition Error: New Dataset and Effective Solution

Jiaxin Zhang, Daizhen Jia, Chongyu Liu, Dezhi Peng, Bangdong Chen, Xue Gao and Lianwen Jin*

IEEE TIM, 2025

Paper / Code

UPOCR: Towards Unified Pixel-Level OCR Interface

D. Peng, Z. Yang, Jiaxin Zhang, C. Liu, Y. Shi, K. Ding, F. Guo, L. Jin

ICML, 2024

Paper / Code

LEGO: Self-Supervised Representation Learning for Scene Text Images

Y. Ren, Jiaxin Zhang, L. Jin

arXiv 2024

Paper

M6Doc: A Large-Scale Multi-Format, Multi-Type, Multi-Layout, Multi-Language, Multi-Annotation Category Dataset for Modern Document Layout Analysis

H. Cheng, P. Zhang, S. Wu, Jiaxin Zhang, Q. Zhu, Z. Xie, J. Li, K. Ding, L. Jin

CVPR, 2023

Paper / Code

Scene Table Structure Recognition with Segmentation Collaboration and Alignment

H. Wang, Y. Xue, Jiaxin Zhang, L. Jin

Pattern Recognition Letters, 165, 146–153, 2023

Paper

Complex Table Structure Recognition in the Wild Using Transformer and Identity Matrix-Based Augmentation

B. Chen, D. Peng, Jiaxin Zhang, Y. Ren, L. Jin

ICFHR, 545–561, 2022

Paper

Multi-scale scene text segmentation examples

Looking from a Higher-Level Perspective: Attention and Recognition Enhanced Multi-Scale Scene Text Segmentation

Y. Ren, Jiaxin Zhang, B. Chen, X. Zhang, L. Jin

ACCV, 643–659, 2022

Paper

TextSRNet: Scene Text Super-Resolution Based on Contour Prior and Atrous Convolution

J. Ma, L. Jin, Jiaxin Zhang, J. Jiang, Y. Xue, M. He

ICPR, 3252–3258, 2022

Paper

Forgery-Free Signature Verification with Stroke-Aware Cycle-Consistent Generative Adversarial Network

J. Jiang, S. Lai, L. Jin, Y. Zhu, Jiaxin Zhang, B. Chen

Neurocomputing, 507, 345–357, 2022

Paper

Academic Service

Conference Reviewer: CVPR, ICCV, NeurIPS, AAAI, ACM MM, SIGGRAPH, ACL, ARR, WACV
Journal Reviewer: TIP, TMM, TCSVT, Pattern Recognition, ACM TOMM