Björn Michele
Ph.D. Student
valeo.ai & IRISA OBELIX
About
I am a Ph.D. student at the valeo.ai and the IRISA OBELIX lab, working on domain adaptation for 3D data. My Ph.D. advisor is Prof. Nicolas Courty who co-advises me with Alexandre Boulch. I also collaborate closely with Gilles Puy, Tuan-Hung Vu, and Renaud Marlet in my research.
I graduated from the Institute Polytechnique Paris with a Master's degree and a thesis on "Zero-shot learning for 3D data". My final year was the M2 Mathematics, Vision, and Apprenticeship (MVA) after completing the Computer Science M1 at the Ecole Polytechnique. I earned my Bachelor's degree from the Baden-Württemberg Cooperative State University in a joint study program with the Robert Bosch GmbH.
If you are looking for a comprehensive overview of publications on domain adaptation and domain generalization for 3D data, check out my curated list: Awesome Domain Adaptation and Generalization for 3D.
Research interests: Computer vision, frugal learning and multimodality.
Recent publications
(2025).
Improving Multimodal Distillation for 3D Semantic Segmentation under Domain Shift.
BMVC, 2025.
(2024).
Train Till You Drop: Towards Stable and Robust Source-free Unsupervised 3D Domain Adaptation.
ECCV, 2024.
(2024).
Horospherical Learning with Smart Prototypes.
Oral @ BMVC, 2024.
(2024).
SALUDA: Surface-based Automotive Lidar Unsupervised Domain Adaptation.
Spotlight @ 3DV, 2024.
(2023).
ALSO: Automotive Lidar Self-supervision by Occupancy estimation.
CVPR, 2023.
Experience
Junior Research Scientist
Education
Ph.D. Computer Science
Thesis: Domain adaptation for 3D data. Expected defense: October 2025.
M2 Mathematics, Vision, and Apprenticeship (MVA)
Thesis: Zero shot learning for 3D data.
M1 Computer Science
BEng Electric Engineering
Thesis: Visualisation of neural networks for plant classification for evaluation and developing purposes.