-
Datasets: CK+ database: You can go to the official website to download
Emotion_labels.zip
andextend-cohn-kanade-images.zip
. You also can go to the Baidu network disk to download datasets. -
Environment needed: Python 3, Tensorflow-gpu, NumPy, opencv, PIL
-
This project is mainly implemented by CNN, which is set-up by myself through tensorflow. In this project, firstly I crop out the main part of face in each picture, and then I save all pictures' data and their labels to csv files. Lastly I load all images' data from csv files as the CNN input. What's more, I saved the weights, losses and accuracies by tensorboard in training process.
jackandsnow / faceexpressionrecognition Goto Github PK
View Code? Open in Web Editor NEWFace Expression Recognition with Deep Learning Method(CNN) through Python 3.