Yu-Xiong Wang

A new approach enhances 3D shape recognition using multi-modal learning techniques.

A new model improves how robots understand their environment in 3D.

A new approach to video object segmentation enhances accuracy by limiting memory use.

A new method improves 3D object alignment and shadow generation from single images.

This study assesses various visual models for understanding complex 3D scenes.

New framework merges image generation and understanding using diffusion models.

Discover RandAR, a new way to create images that breaks traditional boundaries.

Create stunning 4D scenes from simple text prompts with PaintScene4D.

Discover a new method for creating visual programs quickly and cheaply.