Athena Seminar Series: Adapting 3D Reconstruction Models on the Fly, From Test-Time Adaptation to Continual Learning

The NSF AI Institute for Edge Computing (Athena) is pleased to present the next in the Seminar Series by Patrick Rim, titled “Adapting 3D Reconstruction Models on the Fly: From Test-Time Adaptation to Continual Learning” on Wednesday, September 17, 2025, from 1-2pm EST in-person at Yale University (17 Hillhouse, Room 233), and via Zoom. Refreshments will be served.
Biography:
Patrick Rim is a second-year Ph.D. student at the Yale Vision Lab researching 3D vision and multimodal AI systems, advised by Prof. Alex Wong. Previously, he completed his B.S. in Computer Science and Information/Data Sciences at the California Institute of Technology (Caltech). He is also currently a Research Scientist Intern at Meta Reality Labs, working on the Extended Reality (XR) team with Kun He. His work is centered on building embodied AI agents with adaptive, efficient, and robust perception, as well as multimodal capabilities spanning vision, language, and range sensing. To that end, he develops methods for recognition (e.g., multi-sensor 3D object detection in autonomous driving and AR/VR settings), reconstruction (e.g., forecasting dynamic scenes by learning latent ODEs that temporally extrapolate deformable 3D Gaussian splats), and generation (e.g., text-conditional depth map generation with diffusion models). Currently, he is exploring in-the-wild egocentric hand-object tracking to enable contextualized input in unconstrained real-world environments.