Xinyu Wei (魏 心宇)

News

• [2026.06] 🔥🔥🔥LocateAnything surpassed 100K+ Hugging Face downloads within one week of release and 🚀 topped the HF model trend! Have a try!
• [2026.02] One paper accepted by CVPR 2026 🎉
• [2025.09] Two papers accepted by NeurIPS 2025 🎉
• [2025.06] I graduated from Peking University 🎉
• [2025.01] Three papers accepted by ICLR 2025 🎉
• [2024.03] One paper accepted by CVPR 2024 🎉, One paper accepted by CVPR Workshop 💪

Education

• [2022-2025] M.S. in Software Engineering as Merit Student 🎉, Peking University, Supervised by Prof. Shanghang Zhang
• [2018-2022] B.E. in Computer Science with Distinction 🎉, Wuhan University

🔍 Fine-Grained Understanding

As (Co)First-Author

• PAM: Perceive Anything - Recognize, Explain, Caption, and Segment Anything in Images and Videos
  Weifeng Lin*, Xinyu Wei*, Ruichuan An*, Tianhe Ren, Tingwei Chen, Renrui Zhang, Ziyu Guo, Wentao Zhang, Lei Zhang, Hongsheng Li
  Region-level fine-grained understanding with arbitrary kind of visual prompts: segment, recognize, explain, and caption in images and videos.
  ✅ NeurIPS 2025
  [Paper] [Project Page] [Code 🌟279]
• MAVIS: Mathematical Visual Instruction Tuning with an Automatic Data Engine
  Renrui Zhang*, Xinyu Wei*, Dongzhi Jiang, Ziyu Guo, Shicheng Li, Yichi Zhang, Chengzhuo Tong, Jiaming Liu, Aojun Zhou, Bin Wei, Shanghang Zhang, Peng Gao, Chunyuan Li, Hongsheng Li
  The first specialized VLM for multimodal math problem-solving (CLIP-Math + CoT SFT + DPO), with automatic focus on key regions in mathematical figures.
  ✅ ICLR 2025
  [Paper]
• Draw-and-Understand: Leveraging Visual Prompts to Enable MLLMs to Comprehend What You Want
  Weifeng Lin*, Xinyu Wei*, Ruichuan An*, Peng Gao, Bocheng Zou, Yulin Luo, Siyuan Huang, Shanghang Zhang, Hongsheng Li
  Enabling MLLMs to interpret visual prompts (points, boxes, shapes) for fine-grained image comprehension.
  ✅ ICLR 2025
  [Paper] [Project Page] [Code 🌟96]

🎨 Fine-Grained Generation

As (Co)First-Author

• MICo-150K: A Comprehensive Dataset Advancing Multi-Image Composition
  Xinyu Wei*, Kangrui Cen*, Hongyang Wei, Zhen Guo, Bairui Li, Zeqing Wang, Jinrui Zhang, Lei Zhang
  Id-consistent multi-reference image generation demanding fine-grained reference understanding: a dataset, benchmark, and strong baseline model Qwen-MICo.
  ✅ CVPR 2026
  [Paper] [Project Page] [Code 🌟105]
• TIIF-Bench: How Does Your T2I Model Follow Your Instructions?
  Xinyu Wei, Jinrui Zhang, Zeqing Wang, Hongyang Wei, Zhen Guo, Lei Zhang
  Benchmark to systematically assess modern T2I models' ability in following intricate textual instructions.
  Under Review
  [Paper] [Project Page] [Code 🌟128]
• VideoVerse: How Far is Your T2V Generator from a World Model?
  Zeqing Wang*, Xinyu Wei*, Bairui Li, Zhen Guo, Jinrui Zhang, Hongyang Wei, Keze Wang, Lei Zhang
  Benchmark evaluating modern T2V models on temporal causality and world knowledge.
  Under Review
  [Paper] [Project Page] [Code 🌟15]
• PixWizard: Versatile Image-to-Image Visual Assistant with Open-Language Instructions
  Weifeng Lin*, Xinyu Wei*, Renrui Zhang*, Le Zhuo, Shitian Zhao, Siyuan Huang, Huan Teng, Junlin Xie, Yu Qiao, Peng Gao, Hongsheng Li
  Unified image-to-image assistant for generation, manipulation, and translation via free-form language instructions.
  ✅ ICLR 2025
  [Paper] [Code 🌟210]

Other Publications

• GENIUS: Generative Fluid Intelligence Evaluation Suite
  Ruichuan An, Sihan Yang, Ziyu Guo, Wei Dai, Zijun Shen, Haodong Li, Renrui Zhang, Xinyu Wei, Guopeng Li, Wenshan Wu, Wentao Zhang
  Benchmark for evaluating generative fluid intelligence: inducing patterns, executing constraints, and adapting to novel scenarios.
  Under Review
  [Paper]
• UniRef-Image-Edit: Towards Scalable and Consistent Multi-Reference Image Editing
  Hongyang Wei, Bin Wen, Yancheng Long, Yankai Yang, Yuhang Hu, Tianke Zhang, Wei Chen, Haonan Fan, Kaiyu Jiang, Jiankang Chen, Changyi Liu, Kaiyu Tang, Haojie Ding, Xiao Yang, Jia Sun, Huaiqing Wang, Zhenyu Yang, Xinyu Wei, Xianglong He, Yangguang Li, Fan Yang, Tingting Gao, Lei Zhang, Guorui Zhou, Han Li
  Unified framework for single-image editing and multi-image composition with scalable and consistent multi-reference inputs.
  Under Review
  [Paper]
• Delving into RL for Image Generation with CoT: A Study on DPO vs. GRPO
  Chengzhuo Tong, Ziyu Guo, Renrui Zhang, Wenyu Shan, Xinyu Wei, Zhenghao Xing, Hongsheng Li, Pheng-Ann Heng
  First comprehensive study of DPO vs. GRPO in autoregressive image generation with CoT reasoning.
  ✅ NeurIPS 2025
  [Paper] [Code 🌟863]
• Cloud-Device Collaborative Learning for Multimodal Large Language Models
  Guanqun Wang, Jiaming Liu, Chenxuan Li, Junpeng Ma, Yuan Zhang, Xinyu Wei, Kevin Zhang, Maurice Chong, Ray Zhang, Yijiang Liu, Shanghang Zhang
  Enhancing compressed device-deployed MLLMs via cloud collaboration and adapter-based knowledge distillation.
  ✅ CVPR 2024
  [Paper]
• IDAdapter: Learning Mixed Features for Tuning-Free Personalization of Text-to-Image Models
  Siying Cui, Jia Guo, Xiang An, Jiankang Deng, Yongle Zhao, Xinyu Wei, Ziyong Feng
  Tuning-free personalization for text-to-image models from a single face image with identity preservation.
  ✅ CVPR 2024 Workshop
  [Paper]

Hobbies

Photography 📸, Body Building 💪, Movie 🎬, Basketball 🏀, Video Games 🎮

I read history and philosophy 📖

I travel all around the world 🌏

AI is only a brief spark in modern history; modern history is only a blink in agricultural civilization; agricultural civilization is only a thin slice of Homo sapiens' story; and Homo sapiens are only a chapter in the vast history of life. Every age is connected, every phenomenon has a cause.

Service

• Reviewer for ICLR, CVPR, ECCV, NeurIPS
• Cluster & API Quota Administrator, VCLab