This project was from Summer 2018 but I always wanted to improve it. Ain't nobody got time for that so...
You will need 2 sets of images, happy and sad. (If you want to retrain)
`path_happy = "./data/happy/"
path_sad = "./data/sad/"`
And you will need a dataset of the library that I am using (can't remember). https://drive.google.com/file/d/1JocJFmta2eJKVQRC5eRlCX0mzZoUdRVI/view?usp=sharing
WARNING Includes sloppy code, some turkish code that i copypasta'd from another project of mine and was too lazyass to translate, and a free Coca-Cola(no). It works quite well actually
Train:
pythonw train.py -p shape_predictor_68_face_landmarks.dat
Predict:
pythonw predict.py -p shape_predictor_68_face_landmarks.dat -i ./i_am_an_image.jpg