Human Emotion Recognition

Basically, you need to use FER2013 dataset and identify human emotions. Initially, you need to use noise filters and reduce noise in images and then there are 4 ways you should extract features and build deep learning models for each of them and compare which features are resulting in better accuracy. Features extraction ways would be through 1. Gabor Features 2. HOG Features 3. Spatial Features 4. LBP

one thing to keep in mind ... there are readily available projects where we can feed the images directly to model like VGG16 but we should not do that ... we should first remove noise and then extract features and we need to feed those features to model