· Contributors · Organizations · Search
Subdivision Based Convolutional Networks
Research & Education
Full Conference Supporter
Virtual Conference Supporter
Exhibitor Additional Full Conference
Exhibitor Full Conference
DescriptionSubdivNet is an innovative and versatile CNN framework for 3D triangle meshes. It offers a pooling scheme based on the loop subdivision connectivity and mesh face convolution that supports variable kernel size, stride, and dilation. Thereby, many popular 2D CNN architectures can be easily adapted to process 3D meshes.