Neural Network Projects with Python

High accuracy with small data

The third requirement of a facial recognition system is that it needs to be sufficiently accurate (hence secure) while working with a small amount of training data. In Chapter 4, Cats Versus Dogs – Image Classification Using CNNs, we used a huge dataset containing thousands of images of cats and dogs for training our CNN. By contrast, we almost never get this luxury when it comes to the dataset size for facial recognition. Going back to the example of the onboarding process for facial recognition in smartphones, we can see that only a handful of photos are taken, and we need to be able to train our model, using this limited dataset.

Once again, CNNs do not satisfy this requirement, because we need lots of images ...