Face ID recognition

Implemented end-to-end solution for person identification using different models to detect, track, recognize and perform feature-embedding

Face ID has been built end to end for person identification using a deep learning model to detect, track, recognize and embed features. The model is designed by only depending on local descriptors, which can be extracted in an efficient way without the need of any soft or hard constraints. In order to remove false positives that come up from head pose, expression change and aging effects, we have implemented a point-based loss function with multiple training samples.