1 paper
Dequan Kong, Honghua Chen, Zhe Zhu +1
Existing diffusion-based 3D shape completion methods typically use a conditional paradigm, injecting incomplete shape information into the denoising network via deep feature intera…