Zehan Wang

A method for 3D visual grounding using minimal annotations.

New framework improves object location accuracy using relative positioning.

A groundbreaking system merges 3D visuals and language for improved interactions.

Frieren model improves audio quality and sync for video.

OmniBind integrates various data types for improved content understanding and generation.

New methods improve storage and sharing of 3D point clouds.

Learn how 3D models enhance object orientation estimation for tech applications.