test

ShapeFormer: Transformer-based Shape Completion via Sparse Representation

Computer Vision and Pattern Recognition (CVPR), 2022. A transformer-based network that produces a distribution of object completions, conditioned on incomplete, and possibly noisy, point clouds.