How to fine-tune VGG16 for one class classification?

[Alexey Gapchenko]

Hi guys, can you help me please with https://blog.keras.io/building-powerful-image-classification-models-using-very-little-data.html

It is a good example of how to fine-tune conv net for binary classification cats or dogs. what if I want to fine-tune same vgg16 for one class classification?

I want to identify robbers in ski masks I have train/validation small sets. so I need to basically identify 'bank robber' and no bank robber?

[abhishe...@csu.fullerton.edu]

It is a case of Binary classification only.

You have two classes here, 

1. Bank Robber (with ski masks on)

2. No Bank Robber (without ski masks - normal public)