I have been undertaking research concerning on preliminary diagnosis of ADHD children using machine learning. I intend to use facial information such as children's facial expression as the feature. However, it is difficult to get facial expression recognition system with performance over 90%.  

I wonder to feed the face images of ADHD children directly to machine learning and let it recognize the significant pattern that exists among the images. Is there anyone who ever did something like that?

Similar questions and discussions