A New Volumetric CNN for 3D Object Classification Based on Joint Multiscale Feature and Subvolume Supervised Learning Approaches

Figure 5

The network architecture of our proposed MS-VDCNN for an octree depth of 5.