Overall framework
Abstract
IDArb is a diffusion-based method for intrinsic decomposition from an arbitrary number of images under varying illuminations. It uses cross-view, cross-domain attention and illumination-augmented view-adaptive training to estimate multi-view consistent normals and material properties.
Type
Publication
In International Conference on Learning Representations (ICLR)
Tong WU 吴桐
Assistant Professor @ Fudan
My research interests include 3d vision, long-tailed recognition, and robustness.