Shiyang Zhou by the river at night

Shiyang Zhou

周诗洋

Ph.D. Student

HIT Shenzhen
Shenzhen Loop Area Institute

About me

I am a Ph.D. student at Harbin Institute of Technology, Shenzhen, with joint doctoral training at Shenzhen Loop Area Institute (SLAI). I am advised by Prof. Jingyong Su and Prof. Zhuotao Tian.

My research interests are video generation, world models, and world action models. I focus on learning robot behavior from video and improving generalization to new tasks. Previously, I worked on efficient image restoration.

I am looking for research internship opportunities.
Interested in working together? Get in touch.

Experience

Jan 2026 – present

JD · JOY Future Academy

World action models and visual generation.

Apr – Sep 2025

Tencent · Hunyuan

Research Intern · TEG

Data curation and automated pipelines for text-to-image generation and identity-consistent image editing.

Oct 2024 – Apr 2025

OPPO

Research Project

Efficient multi-frame RAW image restoration with Mamba, including joint denoising and demosaicing.

Selected publications

  1. EVO-WAM overview: simulation and real-world robot tasks

    Preprint 2026

    EVO-WAM: Evolving World Action Models through Video-Action Verification

    Shiyang Zhou*, Xionghao Wu*, Wenbo Li, et al.

    Shiyang Zhou*, Xionghao Wu*, Wenbo Li, Shenghe Zheng, Jiyao Zhang, Songsong Yu, Yijun Yang, Jianhui Liu, Haoze Sun, Senqiao Yang, Li Jiang, Jingyong Su, Haoyang Huang, Zhuotao Tian.

    * Equal contribution

    Improving generalization to new tasks through verified imagination, without additional expert demonstrations.

    PaperProjectGitHub
  2. ZimaBlue results: scaling video data improves success and the dual system accelerates control

    Technical report 2026 · Core author

    ZimaBlue: Evolving Generalizable World Action Models through Scalable Video Pre-training

    Xionghao Wu, Yijun Yang, Shiyang Zhou, et al.

    Xionghao Wu, Yijun Yang, Shiyang Zhou, Haoze Sun, Jianhui Liu, Songsong Yu, Jiyao Zhang, Wenbo Li, Bo Wang, Guoqing Ma, Lin Song, Renjie Liao, Shenghe Zheng, Wei Tang, Xiaojuan Qi, Yanwei Li, Yuan Zhang, Zhuotao Tian, Haoyang Huang, Nan Duan.

    Scaling embodied video pre-training for generalizable robot control.

    PaperProjectGitHub
  3. TSANet architecture with Quad-to-Quad and Quad-to-RGB stages

    IEEE TMM 2026

    Lightweight Quad Bayer HybridEVS Demosaicing via State Space Augmented Cross-Attention

    Shiyang Zhou, Haijin Zeng, Yunfan Lu, et al.

    Shiyang Zhou, Haijin Zeng, Yunfan Lu, Yongyong Chen, Jie Liu, Jingyong Su.

    Efficient HybridEVS demosaicing with state space augmented cross-attention.

    Paper
  4. BMTNet architecture combining binarized Mamba and Transformer blocks

    CVPR 2025

    Binarized Mamba-Transformer for Lightweight Quad Bayer HybridEVS Demosaicing

    Shiyang Zhou, Haijin Zeng, Yunfan Lu, et al.

    Shiyang Zhou, Haijin Zeng, Yunfan Lu, Tong Shao, Ke Tang, Yongyong Chen, Jie Liu, Jingyong Su.

    A binarized Mamba-Transformer for lightweight image reconstruction.

    PaperCode

Project contributions

Selected awards

2025Tat-Seng Chua Scholarship
School of Computer Science and Technology, HIT Shenzhen

20243rd Place, CVPR MIPI Challenge
Demosaic for HybridEVS Camera Track

Education

2023 – present

Harbin Institute of Technology, Shenzhen

Ph.D. Student · Computer Science and Technology

2025 – present

Shenzhen Loop Area Institute

Joint Ph.D. Training with HIT Shenzhen

2019 – 2023

Hebei University of Technology

Bachelor’s Degree · Computer Science and Technology

EVO-WAM · Demo

Credits