Xiangyang Xue

New model improves robots' understanding of their surroundings for better task performance.

New framework improves detection of known and unknown objects in three-dimensional space.

EAFormer improves accuracy in isolating text from images using edge information.

Exploring new methods to improve neural surface reconstruction using diverse features.

A new method for seamless 3D editing using multi-view images.

MVGenMaster revolutionizes 3D content creation with innovative image generation techniques.

Robots learn to grasp objects quickly with SparseGrasp, using minimal images.