Paper Reviews Call 002 -- FaceNet: A Unified Embedding for Face Recognition and Clustering

Tess Ferrandez

Notebook: https://github.com/TessFerrandez/rese...

Development, Design, Machine Learning and and sharing what I know to anyone who will listen, that is me in a nutshell...
My focus at the moment is on Machine Learning and specifically the human and sociological aspects of Machine Learning.
Over the last 20 years I have always been interested in patterns, algorithms and computer-human interaction so this is just the culmination of all those interests (oh... and debugging of course) :)

Blog at http://blogs.msdn.microsoft.com/Tess and tweets at @tessferrandez

https://arxiv.org/abs/1503.03832

FaceNet: A Unified Embedding for Face Recognition and Clustering

Florian Schroff, Dmitry Kalenichenko, James Philbin

(Submitted on 12 Mar 2015 (v1), last revised 17 Jun 2015 (this version, v3))

"Despite significant recent advances in the field of face recognition, implementing face verification and recognition efficiently at scale presents serious challenges to current approaches. In this paper we present a system, called FaceNet, that directly learns a mapping from face images to a compact Euclidean space where distances directly correspond to a measure of face similarity. Once this space has been produced, tasks such as face recognition, verification and clustering can be easily implemented using standard techniques with FaceNet embeddings as feature vectors.
Our method uses a deep convolutional network trained to directly optimize the embedding itself, rather than an intermediate bottleneck layer as in previous deep learning approaches. To train, we use triplets of roughly aligned matching / non-matching face patches generated using a novel online triplet mining method. The benefit of our approach is much greater representational efficiency: we achieve state-of-the-art face recognition performance using only 128-bytes per face.
On the widely used Labeled Faces in the Wild (LFW) dataset, our system achieves a new record accuracy of 99.63%. On YouTube Faces DB it achieves 95.12%. Our system cuts the error rate in comparison to the best published result by 30% on both datasets.
We also introduce the concept of harmonic embeddings, and a harmonic triplet loss, which describe different versions of face embeddings (produced by different networks) that are compatible to each other and allow for direct comparison between each other."

COMPLIANCE NOTICE:
THIS WAS FILMED WHILE WE WORKED AT MICROSOFT, NOW I DON'T
EVERYONE IN THIS VIDEO HAS GIVEN PERMISSION TO APPEAR IN THE VIDEO
NO MICROSOFT CONFIDENTIAL INFORMATION WAS DISCUSSED
WE ARE DISCUSSING PUBLIC LITERATURE
THESE ARE OUR PERSONAL OPINIONS
I REFERRED TO THE CHANNEL AS "MACHINE LEARNING AT MICROSOFT", IT HAS NOW BEEN RENAMED "MACHINE LEARNING DOJO" TO MAKE IT CLEAR IT'S NOT AN OFFICIAL MICROSOFT CHANNEL
THIS WAS FILMED IN OUR PERSONAL TIME