Using PyTorch model currently requires the user to define the architecture for some commonly used models. Therefore, we make use of torchvision library to import PyTorch models using labels specifying the name of the model (i.e. resnet50, vgg16) and return layer for feature collection (i.e. avgpool, classifier.4).
Using PyTorch model currently requires the user to define the architecture for some commonly used models. Therefore, we make use of torchvision library to import PyTorch models using labels specifying the name of the model (i.e. resnet50, vgg16) and return layer for feature collection (i.e. avgpool, classifier.4).