Access control with facial recognition

The aim of this work is to create a an access control system using facial recognition. The model used in this project, was trained using triplet loss, specifically with online triplet mining.

Running tests

The model can be found in the project directory called 'model'

The dataset is also included in the project directory path -> test_model\Extracted Faces

The directory test_model -> dataset contains the pictures of two individuals that will be used in testing.

test_image_1 -> Person 1

test_image_2 -> Person 2

To use the model (saved in model), run this line:

py use_model.py 'filename'

for example :

py use_model.py test_image_1.jpg

This will compare the input image 'test_image_1.jpg' to all images of the two persons in dataset directory. The result is expected to be Person 1.

To test the model use :

py test_model.py

Depenedencies

  pip install requirements.txt

RaspberryPI deployment

The script 'use_tflite.py' is the script that will run on the RaspberryPI, using the tensorflow lite model. It works in the same was as the use_model.py script.