3DCSCN: 3D Cascade Shape Completion Network Based On Single Depth View

Depth images of objects can be easily obtained by depth cameras, but they can only provide limited shape information. Current widely learning-based methods generate complete 3D shapes from images, but reconstructed 3D models have low resolutions and noise. To this end, this paper proposes a 3D Cascade Shape Completion Network (3DCSCN) for predicting the complete 3D structure from a single-depth view. 3DCSCN uses an encoder-decoder network to generate rough prediction results and then introduces a point refinement network to update points with high uncertainty for fine-grained prediction results. The experimental results show that 3DCSCN is better than the current methods by an improvement in average IoU and CE by 0.91% and 3.83% on the public ShapeNet dataset, respectively.