Facial expression classification using AlexNet [1]

The AffectNet [2] dataset was used for network training. Τhe classification of images consists of six emotions:

1. Neutral
2. Happy
3. Sad
4. Suprise
5. Anger
6. Fear

References:

[1] Alex Krizhevsky, Ilya Sutskever, Geoffrey E. Hinton, «ImageNet Classification with Deep Convolutional Neural Networks», 2012

[2] Ali Mollahosseini, Behzad Hasani, d Mohammad H. Mahoor, «AffectNet: A Database for Facial Expression, Valence, and Arousal Computing in the Wild», 2017

Examples of correct facial expression prediction:

                       (a) neutral                                                (b) happy                                                  (c) sad

                       (a) suprised                                                (b) anger                                                  (c) fear

Requirements

Python (suggested 3.7.1)  
Numpy   
os-sys  
OpenCv  
Keras (suggested 2.4.3)