In this assignment you will get familiarized with deep learning models, namely CNNs and you will develop a solution for a classification problem, namely the emotion recognition task. Given a dataset of images with different facial expressions labeled as one of the basic emotions (e.g. anger, happiness, surprise, sadness, fear, disgust or neutral), you will need to implement a model based on CNNs, which gets trained on a subset of the dataset and is tested on another set. You will analyze the role and effect of different architectures and parameters of your CNN model (e.g. number of layers, filters, pooling, normalization, etc.), and report your accuracy and observations in a report. Optionally, you can compare your results to the ones achieved using a pre-trained model (e.g. VGG), but the scope of this assignment is to understand the effect of the different CNN operations on your results and apply this model on a real-life sample video.