An innovative way to mark attenence using face recognition.

• Problem Statement
• Solution
• Machine Learning
• Dependecies
• How to run
• Demo
• Project Links

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.

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.

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

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

1. Clone the project
2. Install all the dependencies

pip install -r requirements.txt

3. Save photos in People folder
4. execute while in the directory

python app.py

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

Presentation