UP Lab

members/xiatian.jpeg

Xiatian Zhu, a Senior Lecturer affiliated with the Surrey Institute for People-Centred Artificial Intelligence and the Centre for Vision, Speech and Signal Processing (CVSSP) at the University of Surrey in Guildford, UK, leads the Universal Perception (UP) Lab. Previously a research scientist at Samsung AI Centre, Cambridge, UK, Dr. Zhu holds a Ph.D. from the Queen Mary University of London.

The UP Lab is dedicated to advancing Multimodal Generative AI (GenAI) by modelling diverse data and observations, such as images, videos, audio, text, 3D points, LiDAR, radar, and remote sensing etc. Our applications span a wide range of domains, such as creative media, fashion design, healthcare, climate science, finance, cybersecurity, chemical engineering, sports, gaming, and social issues. Guided by our principle of putting people at the heart of AI, we prioritize the ethical integration of AI capacities with real-world application norms, emphasizing sustainable and responsible AI practices. This ensures our AI initiatives are grounded in community needs and values, aiming to develop transformative technologies that positively impact society and enhance the well-being of individuals and communities.

Email: xiatian.zhu at surrey.ac.uk

News

Mar 1, 2025 Welcome Ms. Yinuo Jing joining us as a visiting student
Feb 28, 2025 Delivered a keynote speech on Autonomous Driving at the 1st Human-Autonomous Vehicle Interaction Workshop, WACV 2025.
Feb 26, 2025 5 CVPR papers accepted
Feb 25, 2025 Serving as Associate Editor for IEEE Transactions on Multimedia
Feb 14, 2025 Welcome applications for Fully Funded PhD Position in Generative AI & Multimodal LLMs (University of Surrey & A*STAR, Singapore)
Feb 10, 2025 Welcome Mr. Yupeng Fang joining us as a visiting student
Feb 3, 2025 Delivered a seminar speech on Multi-modal Generative AI at King's College London.
Jan 27, 2025 Serving as Associate Editor for Neurocomputing
Jan 23, 2025 3 ICLR papers accepted (1 oral with 1.8% acceptance rate, 2 poster)
Jan 15, 2025 Serving as Area Chair for IEEE International Conference on Computer Vision, ICCV 2025
Dec 20, 2024 2 ICASSP papers accepted
Dec 10, 2024 3 AAAI papers accepted
Nov 30, 2024 Area chair, and co-organising a Special Session at ICME 2025: Towards Realistic 3D Deep Learning with Limited Supervision.
Nov 20, 2024 Welcome Dr. Deyin Liu joining us (w/ Dr. Anjan Dutta) as a visiting scholar
Oct 30, 2024 Serving as Associate Editor for Frontiers in Computer Scicence, Computer Vision

Selected publications

  1. CVPR
    FAM Diffusion: Frequency and Attention Modulation for High-Resolution Image Generation with Stable Diffusion
    Yang, Haosen, Bulat, Adrian, Hadji, Isma, Pham, Hai X.,  Zhu, Xiatian, Tzimiropoulos, Georgios, and Martinez, Brais
    In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition 2025
  2. CVPR
    Gaussian Splatting with Localized Points Management
    Yang, Haosen, Zhang, Chenhao, Wang, Wenqing, Volino, Marco, Hilton, Adrian, Zhang, Li, and Zhu, Xiatian
    In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition 2025
  3. CVPR
    Bayesian Test-Time Adaptation for Vision-Language Models
    Zhou, Lihua, Ye, Mao, Li, Shuaifeng, Li, Nianxin,  Zhu, Xiatian, Deng, Lei, Liu, Hongbin, and Lei, Zhen
    In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition 2025
  4. ICLR (Oral, 1.8%)
    Proxy Denoising for Source-Free Domain Adaptation
    Tang, Song, Su, Wenxin, Ye, Mao, Zhang, Jianwei, and Zhu, Xiatian
    In International Conference on Learning Representations 2025
  5. ICLR
    Reflective Gaussian Splatting
    Yao, Yuxuan, Zeng, Zixuan, Gu, Chun,  Zhu, Xiatian, and Zhang, Li
    In International Conference on Learning Representations 2025
  6. ICLR
    SEBRA : Debiasing through Self-Guided Bias Ranking
    Kappiyath, Adarsh, Chaudhuri, Abhra, Jaiswal, Ajay Kumar, Liu, Ziquan, Li, Yunpeng,  Zhu, Xiatian, and Yin, Lu
    In International Conference on Learning Representations 2025
  7. AAAI
    Unsupervised Audio-Visual Segmentation with Modality Alignment
    Bhosale, Swapnil, Yang, Haosen, Kanojia, Diptesh, Deng, Jiangkang, and Zhu, Xiatian
    In AAAI Conference on Artificial Intelligence 2025
  8. AAAI
    OmniCount: Multi-label Object Counting with Semantic-Geometric Priors
    Mondal, Anindya, Nag, Sauradip,  Zhu, Xiatian, and Dutta, Anjan
    In AAAI Conference on Artificial Intelligence 2025
  9. AAAI
    Self-Prompting Analogical Reasoning for UAV Object Detection
    Li, Nianxin, Ye, Mao, Zhou, Lihua, Tang, Song, Gan, Yan, Liang, Zizhuo, and Zhu, Xiatian
    In AAAI Conference on Artificial Intelligence 2025
  10. IJCV
    Preconditioned Score-based Generative Models
    Ma, Hengyuan, Zhang, Li,  Zhu, Xiatian, and Feng, Jianfeng
    International Journal of Computer Vision 2025
  11. NeurIPS
    AV-GS: Learning Material and Geometry Aware Priors for Novel View Acoustic Synthesis
    Bhosale, Swapnil, Yang, Haosen, Kanojia, Diptesh, Deng, Jiankang, and Zhu, Xiatian
    In The Thirty-eighth Annual Conference on Neural Information Processing Systems 2024
  12. NeurIPS (Spotlight, 2.1%)
    Tetrahedron Splatting for 3D Generation
    Gu, Chun, Yang, Zeyu, Pan, Zijie,  Zhu, Xiatian, and Zhang, Li
    In The Thirty-eighth Annual Conference on Neural Information Processing Systems 2024
  13. NeurIPS
    Recognize Any Regions
    Yang, Haosen, Ma, Chuofan, Wen, Bin, Jiang, Yi, Yuan, Zehuan, and Zhu, Xiatian
    In The Thirty-eighth Annual Conference on Neural Information Processing Systems 2024
  14. NeurIPS (Spotlight, 2.1%)
    Motion Forecasting in Continuous Driving
    Song, Nan, Zhang, Bozhou,  Zhu, Xiatian, and Zhang, Li
    In The Thirty-eighth Annual Conference on Neural Information Processing Systems 2024
  15. NeurIPS
    Cloud Object Detector Adaptation by Integrating Different Source Knowledge
    Li, Shuaifeng, Ye, Mao, Zhou, Lihua, Li, Nianxin, Xiao, Siying, Tang, Song, and Zhu, Xiatian
    In The Thirty-eighth Annual Conference on Neural Information Processing Systems 2024