How to train data using VGG-16 with Imagenet model (weight)

Question

I am trying out some code from GitHub. This program is about emotion recognition and I am trying to create bottleneck features to train data using VGG-16 -Imagenet model. I trie