Linyi Jin

University of Michigan–Ann Arbor, Google (United States)

Papers

Total Citations

H-Index

About

Linyi Jin is a researcher whose work sits at the intersection of computer vision and robotics, with a particular focus on enabling machines to understand and interact with complex, cluttered 3D environments. Her key research areas include 3D scene understanding, object manipulation, and dynamic motion estimation from video. Jin’s major contribution lies in addressing the fundamental challenge of occlusion. In her highly cited 2019 and 2022 works on object retrieval from dense clutter, she demonstrated that explicitly inferring the geometry of occluded objects dramatically improves a robot’s ability to locate and grasp a target—a critical capability for real-world applications in warehouse and household automation. More recently, her 2025 work, "Stereo4D," tackles the problem of learning 3D motion from internet stereo videos, proposing a method to supervise dynamic scene understanding at scale without costly ground-truth data. With over 26 citations across her top papers, Jin’s research is steadily gaining recognition for its practical impact. Her work not only advances foundational computer vision but also directly informs more robust and intelligent robotic systems, making her a rising voice in embodied AI.