MLProject: Human Face Expression Recognition

Prerequisite

1. Download FER2013 dataset from kaggle and put train.csv test.csv under ./dataset. Note that you need to manually add label column in test.csv.
2. Then run python generate_data.py, which will create training, validation and test images set under ./dataset/images.

How to run

• SVM: run python svm.py. Refer to the code to change some settings.
• CNN: run python train_cnn.py to train. run python test_cnn.py to test. You can set your own arguments (please refer to the code). demo.py provides a simple demo to predict expression probability given an image.