Lei Wang

About Me

Generative Models · Video Synthesis · Multimodal Learning

I'm a Ph.D. candidate in PCA Lab, VCIP, College of Computer Science, Nankai University. I got my master's degree from the College of Information Science and Technology, Hebei Agricultural University (HEBAU) in 2024.

My master's advisor is Prof. Bo Liu from HEBAU. My Ph.D. advisor is Prof. Jian Yang from Nankai University (NKU), who is a Changjiang Scholar. My vice-advisor is Prof. Yaxing Wang from Jilin University.

I am also a student member of the China Society of Image and Graphics (CSIG).

Outside of academia: I am also trying to live my life with care and intention; at present, I am in a relationship that I deeply cherish and am willing to invest in for the long term. ❤️❤️❤️

Honor

2021–2022 Postgraduate School First Class Scholarships
National Bronze Medal in the “China University Computer Competition - Group Programming Ladder Match” in Hebei Province in 2019.
Second Prize of C/C++ Programming University B Group of the 10th Blue Bridge Cup National Software and Information Technology Professionals Competition Hebei Region.
2018 Third Prize in the Professional Group of the Second College Student Programming Competition of Hebei Agricultural University.

News

Recent academic milestones

2026.06.18: 1 paper (RefAlign) accepted in ECCV 2026.
2026.02.21: 1 paper (WaDi) accepted in CVPR 2026.
2025.02.27: 2 papers (MaskUNet, TiUE) accepted in CVPR 2025.
2023.12.08: I was accepted as a Ph.D. candidate in Computer Science and Technology at the College of Computer Science, Nankai University.
2023.11.17: 1 paper (HiH) submitted in arXiv 2024.
2023.07.14: 1 paper (HSTL) accepted in ICCV 2023.
2023.07.01: 1 paper (GaitMM) accepted in ICIP 2023.
2022.04.04: 1 paper (GaitTB) accepted in 火力与指挥控制.

Publications

* indicates equal contribution, # corresponding author

Featured Publications

RefAlign: Representation Alignment for Reference-to-Video Generation

Lei Wang, Yuxin Song, Ge Wu, Haocheng Feng, Hang Zhou, Jingdong Wang, Yaxing Wang#, Jian Yang#

ECCV, 2026

Paper BibTeX Project Page Code

Key idea: We propose RefAlign, a training-time representation alignment framework that explicitly aligns DiT reference-branch features to the semantic space of a visual foundation model (VFM) to improve identity consistency and reduce cross-modal artifacts.

WaDi: Weight Direction-aware Distillation for One-step Image Synthesis

Lei Wang, Yang Cheng, Senmao Li, Ge Wu, Yaxing Wang#, Jian Yang#

CVPR, 2026

Paper BibTeX Project Page Code

Key idea: We propose the Low-rank Rotation of Weight Direction (LoRaD), a parameter-efficient adapter tailored to one-step diffusion distillation. LoRaD models structured directional changes through learnable low-rank rotation matrices and is integrated into Variational Score Distillation (VSD), resulting in Weight Direction-aware Distillation (WaDi).

Conference Publications

Not All Parameters Matter: Masking Diffusion Models for Enhancing Generation Ability

Lei Wang, Senmao Li, Fei Yang#, Jianye Wang, Ziheng Zhang, Yuhan Liu, Yaxing Wang, Jian Yang#

CVPR, 2025

Paper BibTeX Project Page Code

Key idea: In this paper, we propose a universal masking strategy, MaskUNet, to improve the generation quality and generalization ability of Stable Diffusion.

One-Way Ticket: Time-Independent Unified Encoder for Distilling Text-to-Image Diffusion Models

Senmao Li, Lei Wang, Kai Wang, Tao Liu, Jiehang Xie, Joost van de Weijier, Fahad Shahbaz Khan, Shiqi Yang, Yaxing Wang#, Jian Yang

CVPR, 2025

Paper BibTeX Code

Key idea: We introduce the first Time-independent Unified Encoder (TiUE) architecture, a loop-free distillation approach that eliminates iterative noisy latent processing while maintaining high sampling fidelity with a time cost comparable to previous one-step methods.

Hierarchical Spatio-Temporal Representation Learning for Gait Recognition

Lei Wang, Bo Liu#, Fangfang Liang, Bincheng Wang

ICCV, 2023

Paper BibTeX Code

Key idea: In this paper, we propose a hierarchical spatio-temporal representation learning (HSTL) framework for extracting gait features from coarse to fine.

GaitMM: Multi-Granularity Motion Sequence Learning for Gait Recognition

Lei Wang, Bo Liu#, Bincheng Wang, Fuqiang Yu

ICIP, 2023

Paper arXiv BibTeX Code

Key idea: In this study, we propose a multi-granularity motion representation network (GaitMM) for gait sequence learning.

Preprint

HiH: A Multi-modal Hierarchy in Hierarchy Network for Unconstrained Gait Recognition

Lei Wang, Yinchi Ma, Peng Luan, Wei Yao, Congcong Li, Bo Liu#

arXiv, 2024

Paper BibTeX Code

Key idea: We present a multi-modal Hierarchy in Hierarchy network (HiH) that integrates silhouette and pose sequences for robust gait recognition.

Internships

2025.08 – present, Visual Technology Department, Baidu, Beijing, China. Mentor: Yuxin Song

Review Services

Journal Reviewer

IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI)

Conference Reviewer

IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2024, 2025, 2026
Asian Conference on Computer Vision (ACCV), 2022
Neural Information Processing Systems (NeurIPS), 2026
European Conference on Computer Vision (ECCV), 2026

Friendly Links

Senmao Li (李森茂, NKU) Xiantao Hu (胡现韬, NJUST) Kangneng Zhou (周康能, NKU) Ge Wu (武戈, NKU)