Age, gender, and emotion recognition using deep learning models

The age estimation of a face image can be posed as a deep classification problem using a CNN followed by an expected softmax value refinement (as can be done with a Deep EXpectation (DEX) model). In this recipe, you will first learn how to use a pre-trained deep learning model (a WideResNet with two classification layers added on top of it, which simultaneously estimates the age and gender using a single CNN) for age and gender recognition from a face image. We will use face images from the celebrity faces dataset for age and gender recognition. You will then implement emotion recognition using yet another pre-trained deep learning model, but this time you will need to detect the faces using a face detector (you could use transfer learning, too, and use your classifier on your own images, but this is left as an exercise...