This is a complement version of ‘pseudo-3d-pytorch’ realized by qijiezhao.
- Aiming at training and testing P3D model using my own dataset,I added a few parts including 1.Data Fetching Cell, 2.Training Cell,and3.Evaling Cell.
- The input of P3D model should be preprocessing,more details about preprocessing dataset : https://github.com/yfxc/pseudo-3d-tensorflow (You can add data augmentation cell with refering to 'DataAugmenter.py')
- Using Pytorch,remember executing model.train() when you train model,while model.eval() for testing,especially when you have BatchNorm Layers.