Moving Towards Realistic Symmetry-Based Completion

Computer vision applications rely on 3D geometry recognition and point cloud completion. But relying on deep neural networks (DNN) requires large complex data sets that can be hard to collect.