Department of Automation, Tsinghua University
I am a first-year Ph.D. student in the Department of Automation at Tsinghua University, advised by Prof. Jiwen Lu. My research centers on visual generation and multimodal world models — image, video, 4D generation — with a parallel interest in AI-generated content (AIGC) detection and forensics.
On the methodological side, I work on diffusion, autoregressive generation, 3D/4D reconstruction, and pretraining image generation models.
I'm open to collaborations and discussions—feel free to reach out.
Warning
Problem: The current name of your GitHub Pages repository ("Solution: Please consider renaming the repository to "
http://".
However, if the current repository name is intended, you can ignore this message by removing "{% include widgets/debug_repo_name.html %}" in index.html.
Action required
Problem: The current root path of this site is "baseurl ("_config.yml.
Solution: Please set the
baseurl in _config.yml to "Education
Tsinghua UniversityDepartment of AutomationPh.D. StudentSep. 2025 - Present- Tsinghua UniversityDepartment of AutomationB.S. · GPA 3.90/4.00 (top 10%)Sep. 2021 - Jun. 2025
Honors & Awards
-
Beijing Outstanding Graduate (北京市优秀毕业生)2025
-
Tsinghua Outstanding Student Leader2024
-
Tsinghua Academic Excellence Award (3 consecutive years)2022 - 2024
News
Selected Publications (view all )
Geometry-Aware Single-Image 4D Synthesis via Dense Trajectory Generation
Yanran Zhang*, Ziyi Wang*, Wenzhao Zheng#, Zheng Zhu, Jie Zhou, Jiwen Lu (* equal contribution, # corresponding author)
European Conference on Computer Vision (ECCV) 2026
We propose MoGe4D, a framework that tightly couples 3D geometry reconstruction with motion generation to synthesize dynamic 4D scenes from a single image, together with the TrajScene-60K dataset of dense 4D point trajectories and a diffusion-based 4D trajectory generator.
Geometry-Aware Single-Image 4D Synthesis via Dense Trajectory Generation
Yanran Zhang*, Ziyi Wang*, Wenzhao Zheng#, Zheng Zhu, Jie Zhou, Jiwen Lu (* equal contribution, # corresponding author)
European Conference on Computer Vision (ECCV) 2026
We propose MoGe4D, a framework that tightly couples 3D geometry reconstruction with motion generation to synthesize dynamic 4D scenes from a single image, together with the TrajScene-60K dataset of dense 4D point trajectories and a diffusion-based 4D trajectory generator.
UniGenDet: A Unified Generative-Discriminative Framework for Co-Evolutionary Image Generation and Generated Image Detection
Yanran Zhang, Wenzhao Zheng#, Yifei Li, Bingyao Yu, Yu Zheng, Lei Chen, Jie Zhou, Jiwen Lu (# corresponding author)
IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) 2026
We unify image generation and generated-image detection within a single architecture, where a symbiotic multi-modal attention mechanism and a detector-informed alignment objective allow the two tasks to improve each other in a co-evolutionary loop.
UniGenDet: A Unified Generative-Discriminative Framework for Co-Evolutionary Image Generation and Generated Image Detection
Yanran Zhang, Wenzhao Zheng#, Yifei Li, Bingyao Yu, Yu Zheng, Lei Chen, Jie Zhou, Jiwen Lu (# corresponding author)
IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) 2026
We unify image generation and generated-image detection within a single architecture, where a symbiotic multi-modal attention mechanism and a detector-informed alignment objective allow the two tasks to improve each other in a co-evolutionary loop.
Skyra: AI-Generated Video Detection via Grounded Artifact Reasoning
Yifei Li, Wenzhao Zheng#, Yanran Zhang, Runze Sun, Yu Zheng, Lei Chen, Jie Zhou, Jiwen Lu (# corresponding author)
IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) 2026
We present Skyra, a multimodal large language model that detects AI-generated videos by grounding its decisions in human-perceivable visual artifacts, enabled by the ViF-CoT-4K dataset and a two-stage SFT + reinforcement learning training strategy.
Skyra: AI-Generated Video Detection via Grounded Artifact Reasoning
Yifei Li, Wenzhao Zheng#, Yanran Zhang, Runze Sun, Yu Zheng, Lei Chen, Jie Zhou, Jiwen Lu (# corresponding author)
IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) 2026
We present Skyra, a multimodal large language model that detects AI-generated videos by grounding its decisions in human-perceivable visual artifacts, enabled by the ViF-CoT-4K dataset and a two-stage SFT + reinforcement learning training strategy.
D³QE: Learning Discrete Distribution Discrepancy-aware Quantization Error for Autoregressive-Generated Image Detection
Yanran Zhang, Bingyao Yu, Yu Zheng, Wenzhao Zheng#, Yueqi Duan, Lei Chen, Jie Zhou, Jiwen Lu (# corresponding author)
IEEE/CVF International Conference on Computer Vision (ICCV) 2025
We detect AutoRegressive-generated images by modeling the discrete distribution discrepancy and quantization error in their tokenized representations, validated on a new ARForensics benchmark spanning seven mainstream autoregressive models.
D³QE: Learning Discrete Distribution Discrepancy-aware Quantization Error for Autoregressive-Generated Image Detection
Yanran Zhang, Bingyao Yu, Yu Zheng, Wenzhao Zheng#, Yueqi Duan, Lei Chen, Jie Zhou, Jiwen Lu (# corresponding author)
IEEE/CVF International Conference on Computer Vision (ICCV) 2025
We detect AutoRegressive-generated images by modeling the discrete distribution discrepancy and quantization error in their tokenized representations, validated on a new ARForensics benchmark spanning seven mainstream autoregressive models.
UniPre3D: Unified Pre-training of 3D Point Cloud Models with Cross-Modal Gaussian Splatting
Ziyi Wang*, Yanran Zhang*, Jie Zhou, Jiwen Lu# (* equal contribution, # corresponding author)
IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) 2025
We introduce the first unified pre-training method applicable to 3D point clouds of any scale, predicting Gaussian primitives and rendering via differentiable Gaussian splatting with cross-modal 2D feature guidance.
UniPre3D: Unified Pre-training of 3D Point Cloud Models with Cross-Modal Gaussian Splatting
Ziyi Wang*, Yanran Zhang*, Jie Zhou, Jiwen Lu# (* equal contribution, # corresponding author)
IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) 2025
We introduce the first unified pre-training method applicable to 3D point clouds of any scale, predicting Gaussian primitives and rendering via differentiable Gaussian splatting with cross-modal 2D feature guidance.