Attendance System using Face Recognition

Developed Attendance System using Face Recognition technology to ensure accuracy and speed.

• Software is integrated with the Django web framework and SQLite database to provide a complete package of full-stack python development.
• It identifies a person with an accuracy of 99.38% .
• The model initially starts with 2 training image per person with a prediction probability of 50% over 200 classes using SVM classifier .
• Real-time addition of pictures of the person to the training dataset to improve its performance and accuracy.

Screenshots

Home Page

Prediction with 10 images in training dataset

Admin main page

Statistics Page

Clusters formed Using TSNE for 27 people

Prediction with 300 images at in training dataset in tilted position

Clusters formed Using TSNE for 6 people

This project is a POC application demonstrating the use of facial recognition for marking attendance built as a part of my PS -1 internship at ViitorCloud Technologies, Ahmedabad.

It is a web application that can be used by the company to manage attendance of its employees. It supports admin and employee login. Admin is allowed to register new employees, add their photos to the training dataset, train the model and view attendance reports of all employees. Attendance can be filtered by date or employee. Employee is allowed to view his own attendance reports once he logs in.