In this repo I implemented a simple CNN for face emotion recognition
I used FER2013 dataset which is one of the most challanging datasets that contains approximately 30,000 , 48×48 images in 7 categories :
Angry = 0
Disgust = 1
Fear = 2
Happy = 3
Sad = 4
Surprise = 5
Neutral = 6
you can find the dataset here
the State of art of this dataset is around 75% accuracy and also winner model had 71% accuracy and I reached 63% accuracy in train and validation sets and around 62.7% accuracy in test set in 50 epochs. it is considerable that human-level accuracy of this dataset is only at 65±5% that shows our model has reached to the human level performance.
in this picture you can see the model architecture :
