Deep Learning with PyTorch

Extracting the image features

Here, we combine all the logic that we have seen individually for the algorithms in the chapter:

### For ResNettrn_labels = []trn_resnet_features = []for d,la in train_loader:    o = my_resnet(Variable(d.cuda()))    o = o.view(o.size(0),-1)    trn_labels.extend(la)    trn_resnet_features.extend(o.cpu().data)val_labels = []val_resnet_features = []for d,la in val_loader:    o = my_resnet(Variable(d.cuda()))    o = o.view(o.size(0),-1)    val_labels.extend(la)    val_resnet_features.extend(o.cpu().data)### For Inceptiontrn_inception_features = LayerActivations(my_inception.Mixed_7c)for da,la in train_loader:    _ = my_inception(Variable(da.cuda()))trn_inception_features.remove()val_inception_features = LayerActivations(my_inception.Mixed_7c) ...