????, ????, ????, ???? and ????. A deep convolution neural network was built to train on a "tiny" sample set (images) with five classes (five emotions). Pre-trained classifiers were used to simplify the model.