Created on Sat Aug 7 13:48:22 2021

@author: Seyed Muhammad Hossein Mousavi

Comments: TensorFlow (library)+ Keras (library) + Python (language)+ Spyder (cross-platform integrated development environment) are used to classify 3 facial expressions classes of IKFDB (RGB-D) face dataset [1] using Convolutional Neural Network (CNN) deep learning. Please cite [1] paper if you used the code. Number of samples is 1500 which each class contains 500 ‘jpg’ images. Main folder of ‘deeplr’ contains three subfolders of ‘Joy’, ‘Neutral’ and ‘Sadness’. Custom dataset: You can easily replace your desire dataset here in the main folder and put each class in a separated subfolder. Also, you can increase or decrease number of classes. Of course, you have to change ‘num_classes = 3’ argument according to your number of classes. Please feel free to contact me for any issues or guide: Seyed Muhammad Hossein Mousavi mosavi.a.i.buali@gmail.com [1] Mousavi, S.M.H., Mirinezhad, S.Y. Iranian kinect face database (IKFDB): a color-depth based face database collected by kinect v.2 sensor. SN Appl. Sci. 3, 19 (2021). https://doi.org/10.1007/s42452-020-03999-y