Poor Man's Rekognition

Poor Man’s Rekognition (PMR) is aimed to provide an Open Source alternative to the paid Amazon Rekognition API. This summer, I aim to initiate PMR project providing complete implementation of facial analytics and recognition as a Node.js library. A REST API will also be developed as demonstrated in my Proof of Concept. Node.js library will be a binding for C++ backed Machine Learning (ML) algorithms. OpenFace 2.0 C++ library will be used for face detection and analytics. Tensorflow will be used to implement ArcFace for face recognition. It will be converted to Tensorflowjs model and included in PMR Node.js library. Finally, support for videos will be added exploiting a Multi-Object Tracking algorithm for efficient face attribute and person annotation.