CMU FLAME Center Seminar - Neehar Peri

— 2:00pm

Location:
In Person and Virtual - ET - Tepper Builing 1403

Speaker:
NEEHAR PERI, Ph.D. Student, Robotics Institute , Carnegie Mellon University
https://www.neeharperi.com/

Towards Foundation Models for 3D Perception

Foundation models trained on internet-scale multi-modal data achieves remarkable zero-shot accuracy across a variety of 2D perception tasks. However, extending such foundation models for 3D perception remains challenging due to a lack of diverse large-scale 3D training data. To bridge this gap, we leverage 2D foundation models to generate 3D pseudo-labels, allowing us to turn any 2D dataset into a 3D dataset! We demonstrate the effectiveness of our shelf-supervised approach on popular Autonomous Vehicle (AV) benchmarks across several 3D perception tasks, achieving competitive performance with fully-supervised baselines. Lastly, we highlight applications beyond AVs and examine the limitations of adapting off-the-shelf 2D foundation models for 3D perception in-the-wild. 

— 

Neehar Peri is a fourth year Ph.D. student in the Robotics Institute (RI) at CMU, advised by Prof. Deva Ramanan. He is broadly interested in computer vision and machine learning, with applications to robot perception and action. His current research focuses on leveraging 2D priors to build foundation models for 3D perception. 

In Person and Zoom Participation.  See announcement.

Event Website:
https://www.cmu.edu/flame/events/index.html


Add event to Google
Add event to iCal