SPCNet: Stepwise Point Cloud Completion Network

How will you repair a physical object with large missings? You may first recover its global yet coarse shape and stepwise increase local details. We are motivated to imitate the above procedure address point cloud completion task. propose novel network (SPCNet) for various 3D models missings. SPCNet has hierarchical bottom-to-up architecture. It fulfills in an iterative manner, which 1) infers feature of result; 2) then aid feature; 3) finally detailed result help result. Beyond wisdom simulating repair, we newly design cycle loss enhance generalization robustness SPCNet. Extensive experiments clearly show superiority our over state-of-the-art methods on clouds Code is available at https://github.com/1127368546/SPCNet.