Jiaye Fu — Homepage

About

I am a third-year M.Sc. student at the School of Electronic and Computer Engineering, Peking University. I am fortunate to be co-supervised by Prof. Siwei Ma of the Video Coding Lab and Prof. Jian Zhang of the VILLA Lab. I received my B.Eng. in 2024 from Beijing University of Technology (joint program with University College Dublin), advised by Prof. Jin Wang. I maintain close research collaborations with Dr. Jiaqi Zhang, Prof. Yao Yao, and Dr. Feihu Zhang.

I am currently applying for Fall 2027 Ph.D. positions. I warmly welcome potential advisors and collaborators to contact me.

Research Interests

3D AIGC. Driven by the rise of generative models, I explore efficient and controllable 3D generation and editing for immersive content creation. My recent work contributes to the Neural4D series.
3D Reconstruction. As immersive media emerges as a new information medium, high-quality reconstruction becomes a prerequisite for downstream applications. I focus on efficient, high-fidelity 3D reconstruction techniques for real-world scenes.
VLM Post-training. As vision-language models move from general-purpose understanding toward real-world applications, effectively adapting them to specialized domains becomes increasingly important. I focus on domain-specific post-training techniques for VLMs, aiming to enhance their domain knowledge, multimodal reasoning, and task adaptability.
Next Gen. Video Coding. Video is arguably the most important information carrier of the 21st century, and coding is the core technology that makes it deployable at industrial scale. I focus on the design of next-generation high-performance video coding frameworks, with contributions to JVET and AVS.

Publications

^* Equal contribution. ^† Corresponding author.

Preprints

P2 CodecArena: Codec Quality Assessment via Visual Reinforcement Learning [arXiv] [Project Page] J. Fu, W. Li^*, Q. Gao, Y. Zhao, X. Meng, J. Zhang, S. Ma, J. Zhang. arXiv preprint, 2026.
P1 Spark3R: Asymmetric Token Reduction Makes Fast Feed-Forward 3D Reconstruction [arXiv] Z. Tang^*, J. Fu^*, Q. Gao, H. Li, Y. Wu, J. Zhang, S. Ma, J. Zhang. arXiv preprint, 2026.

Conference Papers

C6 NTIRE 2026 Challenge on 3D Content Super-Resolution: Methods and Results [PDF] L. Wang et al. CVPR Workshops (CVPRW), 2026. CCF-A
C5 WaveletSR: Blind Super-Resolution for Noisy 3D Gaussian Splatting [PDF] Q. Wang, X. Lu, H. Tang, J. Fu, J. Zhang, S. Ma. CVPR Workshops (CVPRW), 2026. CCF-A
C4 ReCon-GS: Continuum-Preserved Gaussian Streaming for Fast and Compact Reconstruction of Dynamic Scenes [PDF] [Code] J. Fu, Q. Gao, C. Wen, Y. Wu, S. Ma, J. Zhang, J. Zhang. NeurIPS 2025. CCF-A
C3 Enhanced Template-based Intra Mode Derivation with Adaptive Block Vector Replacement [PDF] J. Zhang^*, J. Fu^*†, C. Jia, S. Ma, K. Naser, T. Dumas, S. Puri, M. Radosavljevic. PCS 2025.
C2 Advanced Learning-Based Coding Tools for ECM: Intra Prediction and In-Loop Filtering [PDF] Y. Zhao, J. Fu, Z. Li, Q. Wang, Z. Huang, J. Zhang, C. Jia, S. Ma. ISCAS 2025. CCF-B
C1 Recurrent Intra Prediction Mode for Future Video Coding [PDF] J. Fu, X. Meng, S. Ma, J. Zhang, Y. J. Chang, V. Seregin, M. Karczewicz. DCC 2025, pp. 183–192. CCF-B

Journal Papers

J5 MIKU: Material Invariance and Knowledge Unification for PBR Generation Y. Bao, J. Fu, Y. Lin, S. Wu, J. Qian, F. Zhang, X. Cao, Y. Yao. ACM Transactions on Graphics (TOG), 2026. CCF-A
J4 RAPNet: Rate-Distortion-Aware Video Pre-Processing via One-Step Reinforcement Learning for Video Coding [PDF] Y. Zhao, J. Fu, Z. Huang, J. Zhang, S. Ma. IEEE Transactions on Circuits and Systems for Video Technology (TCSVT), 2026. CCF-B
J3 Cross-View-Consistent Feed-Forward 3D Gaussian Splatting for Novel View Synthesis [PDF] X. Lu, J. Fu, J. Zhang, Z. Song, C. Jia, S. Ma. Pattern Recognition (Elsevier), 2026.
J2 TinySplat: Feedforward Approach for Generating Compact 3D Scene Representation [PDF] Z. Song, J. Fu, J. Zhang, X. Lu, C. Jia, S. Ma, W. Gao. IEEE Transactions on Circuits and Systems for Video Technology (TCSVT), 2026. CCF-B
J1 Learned Image Compression via Local-to-global Cross-component Prior [PDF] W. Duan, J. Fu, C. Cui, J. Wu, L. Song, S. Ma, W. Gao. IEEE Transactions on Multimedia (TMM), 2026. CCF-A

50+ proposals contributed to JVET (H.266/VVC successor, ECM) and AVS. Selected below; Adopted marks those integrated into reference software.

2025

AVS-M9322 A RL-based Post-processing Tool for Video Coding — Macau, Dec 2025 Adopted
JVET-AM0074 EE2-1.6: Intra Merge Mode — Daejeon, Jun 2025 Adopted First author
JVET-AM0138 EE2-1.2: TIMD-BV extension with enhanced IntraTMP merge list — Daejeon, Jun 2025 Adopted First author
JVET-AL0290 Enhanced TIMD with NNIP and Optimized Block Vector Derivation — Teleconf, Mar 2025
JVET-AL0195 Non-EE2: TIMD-BV extension with enhanced IntraTMP merge list — Teleconf, Mar 2025 First author
AVS-M8806 QP-adaptive In-Loop Filter tool — Changsha, Mar 2025 Adopted

2024

JVET-AJ0061 EE2-2.5: TIMD merge mode — Kemer, Nov 2024 Adopted
JVET-AJ0269 Non-EE2: OBIC extension with PDP and advanced sampling strategy — Kemer, Nov 2024 First author
JVET-AJ0149 EE2-2.2: Intra merge mode extension with BV development — Kemer, Nov 2024 First author
AVS-M8473 Neural network-based inter prediction filter — Weihai, Aug 2024 Adopted
JVET-AI0224 EE2-related: Intra merge mode extension with BV improvement — Sapporo, Jul 2024 First author
JVET-AI0104 EE2-2.2: TIMD with enhanced block vectors deployment — Sapporo, Jul 2024
JVET-AI0079 Non-EE2: Improved OBIC with PDP — Sapporo, Jul 2024 First author
AVS-M8360 Report of 400 and lossless video coding in AVS3 — Shaoxing, Jun 2024 Adopted
AVS-M8359 EE1-2: DMVR & BIO based motion vector refreshment — Shaoxing, Jun 2024
JVET-AH0071 EE2-related: IntraTMP extension to TIMD — Rennes, Apr 2024 First author
AVS-M8231 DMVR & BIO based motion vector refreshment — Shenzhen, Mar 2024

2023

AVS-M8064 Region-based Intra Template Matching Prediction — Chengdu, Dec 2023 First author
AVS-M8033 Report of 400 and lossless video coding in AVS3 — Chengdu, Dec 2023 Adopted
JVET-AE0120 Non-EE2: Intra Angular Prediction Extension — Geneva, Jul 2023 First author

Awards & Honors

2026 1st Place, 5th Grand Challenge on Neural Network-based Video Coding, ISCAS. [Certificate]
2026 1st Place, NTIRE 3D Content Super-Resolution Challenge (Track 2: Realistic Degradation), CVPR Workshop. [Certificate]
2026 3rd Place, NTIRE 3D Content Super-Resolution Challenge (Track 1: Bicubic Degradation), CVPR Workshop. [Certificate]
2025 Top 10 Student, National Engineering Research Center of Visual Technology (NERCVT), Peking University.
2025 Merit Student, Peking University (~Top 5%).
2024 Outstanding Undergraduate Thesis, Beijing Municipal (Top 0.37% at Beijing).
2024 Outstanding Graduates of Beijing University of Technology.
2021–23 Academic Excellence Award of Beijing University of Technology (three consecutive years).

Education

M.S. in Computer Science, Peking University, Sep. 2024 ~ Jul. 2027 (est.)
Advisors: Prof. Siwei Ma (IEEE Fellow) & Prof. Jian Zhang
B.Eng. in Electronic and Information Engineering (with honors), Beijing University of Technology, Sep. 2020 ~ Jul. 2024
Advisor: Prof. Jin Wang
B.Eng. in Electronic and Information Engineering (joint programme), University College Dublin, Sep. 2020 ~ Jul. 2024

Experience

Researcher, Neural4D, Sep. 2025 ~ Present
Mentor: Prof. Yao Yao
I am responsible for post-training and distillation in 3D mesh and texture generation frameworks, as well as developing and training their inference and acceleration frameworks.

Academic Service

Journal Reviewer: T-IP, T-MM, T-CSVT, TVCG, IEEE Access, VI, J-VCIR, Pattern Recognition, TOMM, SPJ.
Conference Reviewer: NeurIPS, ICML, ACM MM.
Active Contributor: JVET (H.266/VVC successor, ECM) and AVS standards.

Contact

Email: jyfu@stu.pku.edu.cn
GitHub: @jyfu-vcl
LinkedIn: jiaye-fu
WeChat: QR Code