To create a Convolutional Neural Network based model which will be able to identify the gender and emotion of a person in real time.
For achieving our goal, we need to create two separate models. One which will be trained to classify faces based on gender and the other one which will be able to detect emotion of the face. Both these models will be combined to give a real-time detection and prediction of gender and emotion. We have used Tensorflow with Keras for our model training and OpenCV for real-time face detection and prediction.
As we need to create two models, One for gender and the other for emotion. We required two datasets. The data set chosen for this was FER-2013 for Emotion prediction and UTK-Face Dataset for gender prediction.
(This was as a part of project for course EE769 Introduction to Machine Learning at IIT Bombay)
Please use the following link to access the dataset and saved models and weights https://drive.google.com/drive/folders/1nvuuMy4ywccceZ04TIgaV7b2vuTMMRg5?usp=sharing