elmar999 / face-recognition-using-facenet-model

Face recognition using FaceNet model

Introduction

Face recognition using Neural Network. Neural Network is implemented using Keras functional API, that makes it easy to experiment with different architectures. The input of Neural Network is 128 dimensional vector. Haar Cascade has been used to detect faces.

Model

Pretrained FaceNet model has been used to generate 128-D vector embedding and neural network was built on with two hidden layers to classify names.

Installation

First of all, you need to check if you have all required packages to launch the jupyter notebook. If not then install requirements.

$ git clone https://github.com/Elmar999/face-recognition-using-FaceNet-model.git
$ pip3 install -r requirements.txt

If you want to train with your own custom images, you have to prepare your own dataset in train_images folder.

|-face-recognition-using-FaceNet-model
    |-train_images
        |- Person1
            |- img1
            |- img2
            ....
        |- Person2
            |- img1
            |- img2
            ....
        ....

Result