6C classifier

This is an multi-class classifier that is the summary of my experimentation with multi-class multi-label neural classifiers that is able to distinguish between various face coverings. I also created experimental modified version, which is also multi-label, altough it has been succesfully trained, its accuracy is still under my expectations.

The data used for the training contains around 18.000 images, which I have divided into 6 classes :

• Glasses
• Hand
• Mask
• None
• Other
• Sunglasses

I have put a lot of effort into making sure that the data used for training was fairly resourced, balanced and unbiased. The project was built upon Tensorflow 2.0 and Keras libraries.