Skip to content

An attendance monitoring system using face recognition and liveliness detector.

Notifications You must be signed in to change notification settings

old-school-kid/FaceRec

Repository files navigation

FaceRec

An innovative way to mark attenence using face recognition.

Table of Contents

Problem Statement

Online education, though the need of the hour, has also given rise to a lot of instances of students missing classes and cheating. So, we need a robust and innovative system for attendance marking, which is altered for this particular situation.

Our solution

The images are captured and saved during the registration of each student in a database. A model is trained to differentiate between real live fed video and photographs The model first predicts if it is a real human. If it is a real human face then the siamese network comes into action. The siamese network compares photo with all the photo using cosine similarity. If the similarity is above a certain threshold then the person is marked present and the date and time is also noted This data in turn will be saved as a csv file for future reference.

Machine Learning

  1. The face-recognition package which has more than 99% accuracy detects face and encodes them
  2. This encoding is fed into a siamese network to compare the cosine similarity
  3. A threshold (95%) is set. Faces above this similarity are recognized
  4. A model is trained to classify real live image from photos which checks the liveliness

Dependencies

To run the project locally make sure to have the following dependencies installed on your system.

  1. flask
  2. flask-lgoin
  3. flask SQL Alchemy
  4. OpenCV
  5. pandas
  6. Keras
  7. Tensorflow
  8. face-recognition
  9. pyhton-3.x

Instruction

  1. Clone the project
  2. Install all the dependencies
pip install -r requirements.txt
  1. Save photos in People folder
  2. execute while in the directory
python app.py

Demo

The model recognizes the person and marks his/her attendance while recognizing possible fakes like images. Demo

Important Links

Presentation

Releases

No releases published

Packages

No packages published