We have created a very deep convolutional neural network to extract very high-level features from a face for each person. We provide a large-scale image database of faces from many sources, e.g. web crowdsourcing, our in-house integrated dataset. The database contains a wide variety of breeds. The inference time of our model is 115ms on an intel corei7 6700k processor. Mostly in our solution, we save some images from client's SDK which are selected based on our own algorithm. This feature makes our pairing stronger. In addition, we record the checksum of a video to check for duplicate videos.

Our method has shown promising results with wide variations in Appearance, eg pose, age gaps, skin tone, glass, makeup and beard.