This is a derivative work from original project DeepFace by @serengil. I love his initial idea and thank the author, but I consider the code base worth a lot of cleaning and performance enhancements. At this moment my changed code is diverging too much from original so I decided to remove informations which pertain to @serengil alone and his studies and researches.
This fork is all about performance, cleanness and expresiveness of the code while also removing some anti-patterns.
New informations will be published about improvements achieved.