GSOC project progress

Vaibhav Sharma's picture

Today i was able to run the face detection and LDS based recognition system together. The accuracy seems better than MAD. But I could only get the system working for binary classification as the output of multi class classifier of scikit python library was not very reliable for multi class classification. There were some instances where noisy images were classified with high confidence. I am working on such errors and accuracy of the whole so that system can work for multiple classes also.