It recognizes face at real time as well as in image with very good accuracy. It consists of FaceNet deep learning model and SVM statistical model. Now, it is built on small database, database consists of 8 people(5 bollywood actors, 2 my brothers and me), but it can be extended easily on large no. of people. It can be used in attendance system, person tagging in the photo and in other related applications by making some modification in it.

It shows some lags because my laptop has low configuration.

First, The dataset is created by collecting around 15-25 solo photos of each person in different pose, orientation and lighting conditon.Then, faces are extracted from all these photos by using MTCNN library. After that, these face images convert into embeddings representation by using state of the art Facenet-Keras model. Finally, we run SVM classifier to clasify these embeddings into different persons class.

If we want to add new person in the database in future, we can easily done it by following same procedure as above for 1 person.

click here to show dependencies

Licensed under MIT Licencse

• FaceNet is a face recognition system that was described by Florian Schroff, et al. at Google in their 2015 paper titled “FaceNet: A Unified Embedding for Face Recognition and Clustering”.
• A notable example is "Keras FaceNet by Hiroki Taniai".His project provides a script for converting the Inception ResNet v1 model from TensorFlow to Keras. He also provides a pre-trained Keras model ready for use.Download facenet-keras pre-trained model from "here".