Add novel technique in existing Drowsiness detection implementation

I have implemented Drowsiness detection with Opencv and yolo. Code attached.

I want to include some novel approach in existing code so that I can say Drowsiness detection using xyz technique.

Also the new approach you add should be compatible to run in iOS mobile application

I need to get this done on 13-14th October. If you are ready to work
1. Give me what approach you will implement or add in current code.
2. Is it feasible to add in iOS app
3. You need to help me in running code on my machine if I stuck up