Abstract: The purpose of this study is to explore good practice for training convolutional neural networks (CNNs) with spatiotemporal three-dimensional (3D) kernels. Recently, 3D CNNs in the field of ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results