Facial Recognition based Attendance and Access Control

The Requirement:

During the period of Covid-19, companies wanted to have a contact-free attendance system that used facial recognition to grant access to their employees and track attendance.

How did I solve this?

• On an NVIDIA Jetson Nano that is connected to the office network, we run a frontend and backend
• The frontend is run on a tablet that is connected to the office network with a link. It allows to register any employee by adding a few images of them and their details
• On registering, the backend creates embeddings of the faces and stores them
• The backend runs a facial recognition system that matches anyone to the registered users and triggers a pulse from the Jetson Nano on a match. It is attached to a relay that controls the lock of the door
• Also stores the attendance
• Improved performace by using TensorRT for deployment

Technologies:

• Programming Languages: Python
• ML Models: TensorRT (Face Recognition), PyTorch (Face Detection)
• Web Backend: Flask
• Fronend outsourced

Team:

• Abhimanyu Bellam