Computer Vision, 3D and 4D scene reconstruction, Text-driven image, video, and scene generation/editing, Neural surface reconstruction, Multiple view geometry and Structure from Motion (SfM).
Our lab focuses on advancing computer vision, with an emphasis on modeling the dynamic visual world. We actively work on 3D and 4D scene reconstruction from casual videos, developing methods to extract complex geometry and motion. We also research neural surface reconstruction and multi-view geometry to build accurate representations of our environment from images. Additionally, we utilize diffusion models for text-driven editing of images, videos, and 3D scenes.
