How is feature extraction done from by using pre-activations last CNN layer in VGG-19

vision
2

You could use forward hooks to get the activations without changing the model as described here.