Skip to content

RitikMody/face-recognition-system

Repository files navigation

Face Recognition System

Model Architecture

Password Manager Website

  • User can register with an email address and password.
  • User needs to provide his/her photo.
  • Email activation is necessary.
  • User can login with normal email and password or by using Face login option.
  • User can add passwords of other applications.
  • He/she can update and delete the passwords of other applications.
  • An email will be sent in case you need to view the password.

Encryption of password

  • User’s password is stored in an encrypted format using SHA256 algorithm.
  • The passwords in the password manager are encrypted using AES algorithm.
  • For each user, the application generates a unique key which is then used to encrypt the user’s stored passwords

Facial Recognition using Deep Learning

Dataset

We have used the Labelled Faces in the Wild dataset

It consists of :

  • Classes : 2
  • Samples total :13233
  • People :5828

Siamese Network

Architecture

Model:

  • Input: 2 images of size (96,96,3)
  • Gives face embeddings consisting of a vector of 128 features
  • Find the Euclidean Distance between the two face embeddings
  • We then pass the euclidean distance to a sigmoid activation function which gives us a value between 0 to 1.
  • We then consider the faces same if the output is greater than the Threshold.
  • Threshold is set to 0.75 by us.