GSOC project progress

Vaibhav Sharma's picture

Today i tested my LDS program for multiple classes with face detection system. The accuracy was low. The main reason for low accuracy that I have been able to find till now is the wrong output from my multi class supervised classifier of Python Scikit library. For more than two classes the results from the scikit library on my data becomes unacceptable and it is the main reason i am struggling with accuracy. I would be trying some new techniques of making testing and training data to improve accurracy.