Shuguang Dou

Vision · Learning · Discovery

Let RSI shape the data.
Let agents keep learning.
Shuguang Dou, home. Let RSI shape the data. Let agents keep learning.

Vision · Multimodal learning · Healthcare

About me

I lead multimodal research at the Intelligent Healthcare Lab, Huawei.

My work connects multimodal understanding with efficient and trustworthy computer vision. I am interested in how models can reason across visual information, learn from compact data, and support medical image analysis.

I received my Ph.D. in Computer Science and Technology from Tongji University. During my doctoral studies, I interned at Microsoft Research Asia, working with Xinyang Jiang and Dongsheng Li on infographic understanding and neural architecture search. I also visited Michigan State University, hosted by Prof. Xiaoming Liu. My doctoral research focused on robustness, security, and privacy in person re-identification.

We welcome research scientists, engineers, and interns interested in these directions. Get in touch

Research directions

  • Agentic Vision
  • Model and dataset distillation
  • Medical image analysis

Recent news

All updates
  • Code for PIS (SCIS 2026) and DROP (TBIOM 2026) is now available. See the Code page for the curated implementations.
  • A paper on 'Distilling and Scaling Hierarchical Vision Transformer to 30B Parameters' is accepted by IJCAI 2026.
  • A paper on 'Decouple Re-identification and Human Parsing for Occluded Person Re-Identification' is accepted by IEEE TBIOM in 2026.
  • A paper on 'Agentic Video Summarization via Self-Reflecting Multimodal Understanding' is accepted by CVPR 2026. Congratulations to my intern Miaotian Guo!
  • A paper on 'Active Dataset Distillation via Dual-Space Informative Matching' is accepted by IEEE TIP in 2026.
  • A paper on 'Asynchronous Matching with Dynamic Sampling for Multimodal Dataset Distillation' is accepted by ICLR 2026.
  • I have been invited to be a Reviewer for the ECCV 2026 Conference.
  • I have been invited to be a Reviewer for the ICML 2026 Conference.
  • I have been invited to be a Reviewer for the CVPR 2026 Conference.
  • A paper on 'Visual Bridge:Universal Visual Perception Representations Generating' is accepted by AAAI 2026. Congratulations to my intern Yilin Gao!
  • I have been invited to be a Reviewer for the ICLR 2026 Conference.
  • A paper on 'Person Identity Shift for Privacy-Preserving Person Re-identification' is accepted by SCIS in 2025 (CCF A).

Selected publications

Recent work and representative research.

All publications

IJCAI2026

Distilling and Scaling Hierarchical Vision Transformer to 30B Parameters

Shuguang Dou, Dongqi Li, Hao Jiang, Dongsheng Jiang

Studies distillation and scaling of hierarchical vision transformers to 30 billion parameters.

Scholar: Distilling and Scaling Hierarchical Vision Transformer to 30B Parameters
BibTeX: Distilling and Scaling Hierarchical Vision Transformer to 30B Parameters 
@inproceedings{dou2026hierarchical30b,
  title = {Distilling and Scaling Hierarchical Vision Transformer to 30B Parameters},
  author = {Dou, Shuguang and Li, Dongqi and Jiang, Hao and Jiang, Dongsheng},
  booktitle = {International Joint Conference on Artificial Intelligence},
  year = {2026},
  url = {https://scholar.google.com/citations?view_op=view_citation&hl=zh-CN&user=n2YT06EAAAAJ&citation_for_view=n2YT06EAAAAJ:HDshCWvjkbEC}
}

CVPR2026

Research overview for Agentic Video Summarization via Self-Reflecting Multimodal Understanding

Agentic Video Summarization via Self-Reflecting Multimodal Understanding

Miaotian Guo, Shuguang Dou, Yin Li, Aidong Men, Dongsheng Jiang

A multimodal agent workflow summarizes videos through autonomous verification and reflection.

Paper: Agentic Video Summarization via Self-Reflecting Multimodal Understanding Code: Agentic Video Summarization via Self-Reflecting Multimodal Understanding
BibTeX: Agentic Video Summarization via Self-Reflecting Multimodal Understanding 
@inproceedings{agenticvs2026,
  title = {{Agentic Video Summarization via Self-Reflecting Multimodal Understanding}},
  author = {Miaotian Guo and Shuguang Dou and Yin Li and Aidong Men and Dongsheng Jiang},
  booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  year = {2026},
  pages = {40516--40526},
  url = {https://openaccess.thecvf.com/content/CVPR2026/html/Guo_Agentic_Video_Summarization_via_Self-Reflecting_Multimodal_Understanding_CVPR_2026_paper.html}
}

AAAI2026

Research overview for Visual Bridge: Universal Visual Perception Representations Generating

Visual Bridge: Universal Visual Perception Representations Generating

Yilin Gao, Shuguang Dou, Junzhou Li, Zhiheng Yu, Yin Li, Dongsheng Jiang, Shugong Xu

A universal flow-matching framework generates representations for multiple visual perception tasks.

Paper: Visual Bridge: Universal Visual Perception Representations Generating
BibTeX: Visual Bridge: Universal Visual Perception Representations Generating 
@inproceedings{visualbridge2026,
  title = {{Visual Bridge: Universal Visual Perception Representations Generating}},
  author = {Yilin Gao and Shuguang Dou and Junzhou Li and Zhiheng Yu and Yin Li and Dongsheng Jiang and Shugong Xu},
  booktitle = {Proceedings of the AAAI Conference on Artificial Intelligence},
  year = {2026},
  volume = {40},
  number = {25},
  pages = {21234--21242},
  doi = {10.1609/aaai.v40i25.39268}
}

TBIOM2026

Research overview for Decouple Re-Identification and Human Parsing for Occluded Person Re-Identification

Decouple Re-Identification and Human Parsing for Occluded Person Re-Identification

Shuguang Dou, Yuanpeng Tu, Yanna Lin, Junyao Gao, Cairong Zhao

Decoupled identity and human-parsing features improve person retrieval under occlusion.

Paper: Decouple Re-Identification and Human Parsing for Occluded Person Re-Identification Code: Decouple Re-Identification and Human Parsing for Occluded Person Re-Identification
BibTeX: Decouple Re-Identification and Human Parsing for Occluded Person Re-Identification 
@article{drop2026,
  title = {{Decouple Re-Identification and Human Parsing for Occluded Person Re-Identification}},
  author = {Shuguang Dou and Yuanpeng Tu and Yanna Lin and Junyao Gao and Cairong Zhao},
  journal = {IEEE Transactions on Biometrics, Behavior, and Identity Science},
  year = {2026},
  volume = {8},
  number = {4},
  pages = {666--677},
  doi = {10.1109/TBIOM.2026.3680760}
}

SCIS2026

Research overview for Person identity shift for privacy-preserving person re-identification

Person identity shift for privacy-preserving person re-identification

Shuguang Dou, Xinyang Jiang, Qingsong Zhao, Yansen Wang, Dongsheng Li, Cairong Zhao

Identity shifting protects personal appearance while preserving identity relationships needed for person re-identification.

Paper: Person identity shift for privacy-preserving person re-identification Code: Person identity shift for privacy-preserving person re-identification
BibTeX: Person identity shift for privacy-preserving person re-identification 
@article{pis2026,
  title = {{Person identity shift for privacy-preserving person re-identification}},
  author = {Shuguang Dou and Xinyang Jiang and Qingsong Zhao and Yansen Wang and Dongsheng Li and Cairong Zhao},
  journal = {Science China Information Sciences},
  year = {2026},
  volume = {69},
  number = {1},
  doi = {10.1007/s11432-023-4583-x},
  pages = {112103}
}

TPAMI2025

Research overview for EA-HAS-Bench and Language-Enhanced Shrinkage Search for Energy-Aware NAS

EA-HAS-Bench and Language-Enhanced Shrinkage Search for Energy-Aware NAS

Shuguang Dou, Jiale Zhao, Xinyang Jiang, Junyao Gao, Yuge Zhang, Bo Li, Dongsheng Li

Language-guided search and an energy-aware benchmark support more efficient neural architecture and hyperparameter optimization.

Paper: EA-HAS-Bench and Language-Enhanced Shrinkage Search for Energy-Aware NAS
BibTeX: EA-HAS-Bench and Language-Enhanced Shrinkage Search for Energy-Aware NAS 
@article{less2025,
  title = {{EA-HAS-Bench and Language-Enhanced Shrinkage Search for Energy-Aware NAS}},
  author = {Shuguang Dou and Jiale Zhao and Xinyang Jiang and Junyao Gao and Yuge Zhang and Bo Li and Dongsheng Li},
  journal = {IEEE Transactions on Pattern Analysis and Machine Intelligence},
  year = {2025},
  volume = {47},
  number = {12},
  pages = {11596--11613},
  doi = {10.1109/TPAMI.2025.3598206}
}

TPAMI2024

Research overview for Hierarchically Recognizing Vector Graphics and A New Chart-Based Vector Graphics Dataset

Hierarchically Recognizing Vector Graphics and A New Chart-Based Vector Graphics Dataset

Shuguang Dou, Xinyang Jiang, Lu Liu, Lu Ying, Caihua Shan, Yifei Shen, Xuanyi Dong, Yun Wang, Dongsheng Li, Cairong Zhao

Hierarchical representations recognize objects directly in vector graphics, supported by a chart understanding benchmark.

Paper: Hierarchically Recognizing Vector Graphics and A New Chart-Based Vector Graphics Dataset Code: Hierarchically Recognizing Vector Graphics and A New Chart-Based Vector Graphics Dataset Project: Hierarchically Recognizing Vector Graphics and A New Chart-Based Vector Graphics Dataset
BibTeX: Hierarchically Recognizing Vector Graphics and A New Chart-Based Vector Graphics Dataset 
@article{yolatplusplus2024,
  title = {{Hierarchically Recognizing Vector Graphics and A New Chart-Based Vector Graphics Dataset}},
  author = {Shuguang Dou and Xinyang Jiang and Lu Liu and Lu Ying and Caihua Shan and Yifei Shen and Xuanyi Dong and Yun Wang and Dongsheng Li and Cairong Zhao},
  journal = {IEEE Transactions on Pattern Analysis and Machine Intelligence},
  year = {2024},
  volume = {46},
  number = {12},
  pages = {7556--7573},
  doi = {10.1109/TPAMI.2024.3394298}
}

ICLR2023

Research overview for EA-HAS-Bench: Energy-Aware Hyperparameter and Architecture Search Benchmark

EA-HAS-Bench: Energy-Aware Hyperparameter and Architecture Search Benchmark

Shuguang Dou, Xinyang Jiang, Cairong Zhao, Dongsheng Li

An energy-aware benchmark studies trade-offs between model quality and the cost of architecture and hyperparameter search.

Paper: EA-HAS-Bench: Energy-Aware Hyperparameter and Architecture Search Benchmark PDF: EA-HAS-Bench: Energy-Aware Hyperparameter and Architecture Search Benchmark Code: EA-HAS-Bench: Energy-Aware Hyperparameter and Architecture Search Benchmark Project: EA-HAS-Bench: Energy-Aware Hyperparameter and Architecture Search Benchmark
BibTeX: EA-HAS-Bench: Energy-Aware Hyperparameter and Architecture Search Benchmark 
@inproceedings{eahasbench2023,
  title = {{EA-HAS-Bench: Energy-Aware Hyperparameter and Architecture Search Benchmark}},
  author = {Shuguang Dou and Xinyang Jiang and Cairong Zhao and Dongsheng Li},
  booktitle = {International Conference on Learning Representations},
  year = {2023},
  url = {https://openreview.net/forum?id=n-bvaLSCC78}
}

TIP2023

Research overview for Human Co-Parsing Guided Alignment for Occluded Person Re-Identification

Human Co-Parsing Guided Alignment for Occluded Person Re-Identification

Shuguang Dou, Cairong Zhao, Xinyang Jiang, Shanshan Zhang, Wei-Shi Zheng, Wangmeng Zuo

Weakly supervised human co-parsing aligns visible body regions to improve occluded person re-identification.

Paper: Human Co-Parsing Guided Alignment for Occluded Person Re-Identification PDF: Human Co-Parsing Guided Alignment for Occluded Person Re-Identification Code: Human Co-Parsing Guided Alignment for Occluded Person Re-Identification Project: Human Co-Parsing Guided Alignment for Occluded Person Re-Identification
BibTeX: Human Co-Parsing Guided Alignment for Occluded Person Re-Identification 
@article{hcga2023,
  title = {{Human Co-Parsing Guided Alignment for Occluded Person Re-Identification}},
  author = {Shuguang Dou and Cairong Zhao and Xinyang Jiang and Shanshan Zhang and Wei-Shi Zheng and Wangmeng Zuo},
  journal = {IEEE Transactions on Image Processing},
  year = {2023},
  volume = {32},
  pages = {458--470},
  doi = {10.1109/TIP.2022.3229639}
}
Explore the full publication archive