M. Saquib Sarfraz

Computer Vision, Machine Learning, AI

I am Principal Lead of Deep Learning at Mercedes-Benz, where I drive strategic AI innovation across generative AI, autonomous perception, and smart manufacturing. My work focuses on bringing cutting-edge machine learning and computer vision research into scalable real-world systems for automotive and industrial applications. My research has contributed methodological advances in parameter-free clustering and graph-based structure discovery.. Previously, I served as Senior Scientist and Lecturer at Karlsruhe Institute of Technology (KIT), where I led research on human-centered perception and computer vision for HCI-related applications. I received my PhD in Computer Vision from Technical University of Berlin, Germany in 2009. Over the years, my work has been recognized with several Best Paper Awards and Outstanding Reviewer Awards at leading AI conferences. Earlier in my career, I was Assistant Professor at the Electrical Engineering department of COMSATS Institute of Technology in Lahore, Pakistan, where I founded and directed the Computer Vision Research Group.

Research Interests: Multimodal Large Language Models, Deep Learning, Computer Vision, Image and Video Understanding

news/highlights

May 20, 2026 Invited Talk: ā€œSeeing Domain Generalization in Foundation Modelsā€, DG-EBF at CVPR 2026.
May 1, 2026 paper/s accepted at ICLR 2026, TMLR 2026 and IJCV 2026!
Mar 1, 2026 Serving as Area Chair for ECCV 2026.
Jun 12, 2025 CVPR 2025 tutorial on identifying structure in data
Jun 1, 2025 Outstanding Reviewer Awards at CVPR 2025, 2024, 2023, 2020, 2021 and ICCV 2021 :1st_place_medal:
Apr 17, 2024 Invited Talk at Cohere For AI on Clustering and Dimentionality Reduction Watch on Youtube
Jan 1, 2017 Press/Media coverage on Thermal to Visible Face Recognition for night-time surveillance

selected publications

  1. ICML
    Position: Quo Vadis, Unsupervised Time Series Anomaly Detection?
    Sarfraz, M Saquib, Chen, Mei-Yen, Layer, Lukas, Peng, Kunyu, and Koulakis, Marios
    In International Conference on Machine Learning 2024
    HTML PDF Code
  2. CVPR
    Hierarchical Nearest Neighbor Graph Embedding for Efficient Dimensionality Reduction
    Sarfraz, M Saquib, Koulakis, Marios, Seibold, Constantin, and Stiefelhagen, Rainer
    In IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) 2022
    HTML PDF Code
  3. CVPR
    Temporally-Weighted Hierarchical Clustering for Unsupervised Action Segmentation
    Sarfraz, M Saquib, Murray, Naila, Sharma, Vivek, Diba, Ali, Van Gool, Luc, and Stiefelhagen, Rainer
    In IEEE/CVF International Conference on Computer Vision and Pattern Recognition (CVPR) 2021
    HTML PDF Code
  4. CVPR
    Efficient parameter-free clustering using first neighbor relations
    Sarfraz, M Saquib, Sharma, Vivek, and Stiefelhagen, Rainer
    In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) 2019
    HTML PDF Code
  5. CVPR
    A pose-sensitive embedding for person re-identification with expanded cross neighborhood re-ranking
    Sarfraz, M Saquib, Schumann, Arne, Eberle, Andreas, and Stiefelhagen, Rainer
    In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) 2018
    HTML PDF Code