Profile
I am a Postdoctoral Research Fellow at the State Key Laboratory of Integrated Service Networks, Xidian University. I work on learning-based video compression and visual representation under resource-constrained scenarios, aiming to unify efficient transmission, high-quality reconstruction, and machine understanding. I have published papers in leading conferences and journals, including CVPR, AAAI, ACM MM, IJCAI, IEEE T-CSVT, and IEEE T-IV. I serve as a reviewer for CVPR, AAAI, IJCAI, ACM MM, IEEE T-CSVT, IEEE TNNLS, and IJCV.
Research Interests
Learned compression and variable-rate representation
Image/video enhancement and high-fidelity reconstruction
Bridging human perception and machine understanding
News
- 2026.02: 🎉 Two papers are accepted by CVPR 2026
- 2025.11: 🎉 One paper is accepted by AAAI 2026 as an oral presentation
- 2025.07: 🎉 Received funding from China Postdoctoral Science Foundation under Grant 2025M773501
- 2025.04: 🎉 Two papers are accepted by IJCAI 2025
- 2025.03: 🎉 Received funding from the Fundamental Research Funds for the Central Universities under Grant ZYTS25270
- 2024.07: 🎉 One paper is accepted by IEEE Transactions on Intelligent Vehicles
- 2024.07: 🎉 One paper is accepted by ACM MM 2024
- 2024.06: 🎉 Completed my Ph.D. at Xidian University
- 2024.04: 🎉 One paper is accepted by IJCAI 2024
📝 Selected Publications

RealRep: Generalized SDR-to-HDR Conversion via Attribute-Disentangled Representation Learning
Li Xu, Siqi Wang, Kepeng Xu, Gang He, Lin Zhang, Weiran Wang, Yu-Wing Tai.
- Attribute-disentangled representation learning for more generalizable real-world SDR-to-HDR conversion.

Unleashing the Potential of Transformer Flow for Photorealistic Face Restoration
Kepeng Xu, Li Xu, Gang He, Wei Chen, Xianyun Wu, Wenxin Yu.
- Transformer flow matching for high-quality photorealistic face restoration.

Beyond Feature Mapping GAP: Integrating Real HDRTV Priors for Superior SDRTV-to-HDRTV Conversion
Gang He, Kepeng Xu, Li Xu, Zhiqiang Zhang, Wenxin Yu, Shihao Wang, Dajiang Zhou, Yunsong Li.
- Real HDRTV priors for improved SDR-to-HDR conversion in practical scenarios.


An End-to-End Real-World Camera Imaging Pipeline
Kepeng Xu, Zijia Ma, Li Xu, Gang He, et al.
- End-to-end real-world camera imaging pipeline for complex distortions.

Beyond Alignment: Blind Video Face Restoration via Parsing-Guided Temporal-Coherent Transformer
Kepeng Xu, Li Xu, Gang He, et al.
- Parsing-guided temporal-coherent transformer for clearer and more consistent video face restoration.

SDRTV-to-HDRTV via Hierarchical Dynamic Context Feature Mapping
Gang He, Kepeng Xu, Li Xu, Chang Wu, Ming Sun, Xing Wen, Yu-Wing Tai.
- Hierarchical dynamic context feature mapping for high-quality HDR conversion.

Transcoded Video Restoration by Temporal Spatial Auxiliary Network
Li Xu, Gang He, Jinjia Zhou, Jie Lei, Weiying Xie, Yunsong Li, Yu-Wing Tai.
- Temporal-spatial auxiliary network for improving transcoded video quality.
-
AAAI 2026 OralLi Xu, Siqi Wang, Kepeng Xu, Gang He, Lin Zhang, Weiran Wang, Yu-Wing Tai. RealRep: Generalized SDR-to-HDR Conversion via Attribute-Disentangled Representation Learning [C]. Proceedings of the AAAI Conference on Artificial Intelligence, 2026. -
IJCAI 2025Kepeng Xu, Li Xu, Gang He, Wei Chen, Xianyun Wu, Wenxin Yu. Unleashing the Potential of Transformer Flow for Photorealistic Face Restoration [C]. Proceedings of the Thirty-Fourth International Joint Conference on Artificial Intelligence, Montreal, Canada, 2025. -
IJCAI 2025Gang He, Kepeng Xu, Li Xu, Zhiqiang Zhang, Wenxin Yu, Shihao Wang, Dajiang Zhou, Yunsong Li. Beyond Feature Mapping GAP: Integrating Real HDRTV Priors for Superior SDRTV-to-HDRTV Conversion [C]. Proceedings of the Thirty-Fourth International Joint Conference on Artificial Intelligence, Montreal, Canada, 2025. -
T-IVLi Xu, Shaodi You, Gang He, Yunsong Li. Pedestrian-Vehicle Information Modulation for Pedestrian Crossing Intention Prediction [J]. IEEE Transactions on Intelligent Vehicles. -
ACM MM 2024Kepeng Xu, Zijia Ma, Li Xu, Gang He, Yunsong Li, Wenxin Yu, Taichu Han, Cheng Yang. An End-to-End Real-World Camera Imaging Pipeline, Melbourne, Australia, 2024. -
IJCAI 2024Kepeng Xu, Li Xu, Gang He, Wenxin Yu, Yunsong Li. Beyond Alignment: Blind Video Face Restoration via Parsing-Guided Temporal-Coherent Transformer [C]. Proceedings of the Thirty-Third International Joint Conference on Artificial Intelligence, Jeju, Korea, 2024. -
NCGang He, Shaoyi Long, Li Xu, Chang Wu, Jinjia Zhou, Ming Sun, Yurong Dai. Global priors guided modulation network for joint super-resolution and SDRTV-to-HDRTV [J]. Neurocomputing 554 (2023): 126590. -
ArxivKepeng Xu, Li Xu, Gang He, Wenxin Yu, Yunsong Li. Towards Robust SDRTV-to-HDRTV via Dual Inverse Degradation Network. 2023. -
AAAI 2022Li Xu, Gang He, Jinjia Zhou, Jie Lei, Weiying Xie, Yunsong Li, Yu-Wing Tai. Transcoded Video Restoration by Temporal Spatial Auxiliary Network [C]. Proceedings of the AAAI Conference on Artificial Intelligence, Vancouver, BC, Canada, 2022: Vol. 36. No. 3. 2022. -
DSPLi Xu, Chang Wu, Linyi Huang, Shengyu Wei, Gang He, Yunsong Li. Towards coding for VoD application: An enhanced video compression system with a content-fitted recursive restoration network [J]. Digital Signal Processing 122 (2022): 103368. T-CSVTGang He, Li Xu, Jie Lei, Weiying Xie, Yunsong Li, Yibo Fan, Jinjia Zhou. Interlayer Restoration Deep Neural Network for Scalable High Efficiency Video Coding [J]. IEEE Transactions on Circuits and Systems for Video Technology, vol. 32, no. 5, pp. 3217-3234, May 2022.ACM MM 2022Gang He, Kepeng Xu, Li Xu, Chang Wu, Ming Sun, Xing Wen, Yu-Wing Tai. SDRTV-to-HDRTV via Hierarchical Dynamic Context Feature Mapping [C]. Proceedings of the 30th ACM International Conference on Multimedia, Lisbon, Portugal, 2022. p. 2890-2898.CVPRW 2022Gang He, Yong Wang, Li Xu, Wenli Zhang, Ming Sun, Xing Wen. Focused Feature Differentiation Network for Image Quality Assessment [C]. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshop, New Orleans, Louisiana, USA, 2022. p. 1800-1804.ArxivKepeng Xu, Li Xu, Zijia Ma, Ming Sun, Yu-Wing Tai. Sdrtv-to-hdrtv conversion via spatial-temporal feature fusion. 2022.
🎖 Honors and Awards
- 2022 Chinese National Scholarship for Ph.D. Students
- 2022 The 5th Challenge on Learned Image Compression, CVPR workshop (3rd Place)
- 2020 Outstanding Graduate Student at Xidian University
📖 Education
- 2020.03 - 2024.06, Ph.D. in Information and Communication Engineering, Xidian University, Xi’an, China
- 2022.10 - 2024.01, Guest Researcher, Faculty of Science, University of Amsterdam, Amsterdam, Netherlands
- 2018.09 - 2020.02, M.S. in Information and Communication Engineering, Xidian University, Xi’an, China
- 2014.09 - 2018.06, B.S. in Communication Engineering, Chongqing University, China