Patrick Labatut
Papers
1
Total Citations
3
H-Index
1
About
Patrick Labatut is a leading researcher in self-supervised learning and video understanding, with a focus on building AI systems that learn from observation rather than explicit labels. His most notable contribution is the development of V-JEPA 2, a groundbreaking self-supervised video model that enables machines to understand, predict, and plan actions by combining internet-scale video data with minimal robot interaction data. This work addresses a fundamental challenge in AI: learning world models that can generalize across diverse tasks with little to no supervision. While still early in its citation trajectory, V-JEPA 2 represents a paradigm shift toward more efficient, observation-driven learning. Labatut’s research sits at the intersection of computer vision, robotics, and representation learning, pushing the boundaries of how machines can acquire common sense and physical understanding from passive video. His work is particularly impactful for students and researchers interested in scalable self-supervised methods, video-based learning, and the path toward more autonomous, perceptive AI systems.
Research Focus
Key Achievements
Top Papers
- 1