Facemask Detector Project

This project is the final class project of an AI classes during last year of IT Engineering School.

Goal

The main goal of this project is to develop and train an AI (completely or by using existing ones) that will detect faces with or without masks, and also improperly worn masks.

Datasets

• The evaluation dataset is the Face Mask Detection dataset by andrewmvd on Kaggle.
• The dataset used to train this detector is a custom dataset made by all the groups working on this project.

We will use the same classes as the Kaggle dataset :

• with_mask : the person is wearing a face mask properly,
• mask_weared_incorrect : the person is wearing a face mask improperly,
• without_mask : the person is not wearing a face mask.

AI

We use the Efficientdet D1 and Efficientdet D0 AIs from the Tensorflow Object Detection API that we have trained with our custom dataset.

How to use

1. Download or clone this repository
2. Create a new Python environment (virtualenv, conda...)
3. Using this environment, install pip packages using requirements.txt file
   • In your terminal, type python -m pip install -r requirements.txt,
   • If you have a CUDA-Capable GPU, you can install CUDA and cuDNN for the last TensorFlow version.
4. Start the FaceMaskDetector.py script using same environment.

To start inference on the eval dataset :
python FaceMaskDetector.py --eval [--version <version name>] [--minScore <score>] [--save] [--noviz]

To start inference on the camera stream :
python FaceMaskDetector.py --inference [--version <version name>] [--minScore <score>] [--camera <camera name>]

To start inference on a single image :
python FaceMaskDetector.py --inference [--version <version name>] [--minScore <score>] --file <file path> [--save] [--noviz]

To display the help message :
python FaceMaskDetector.py -h

usage: [-h] [-i] [-v {d0_v1,d1_v1}] [-f FILE_PATH] [-s] [-m MIN_SCORE] [-nv] [-c CAMERA_NAME]

optional arguments:
-h, --help            Show this help message and exit
-i, --inference       Start the Face Mask Detector in Inference mode on camera input or given image. Not adding this flag 
						will start the Face Mask Detector in Evaluation mode (run inferences on "images" folder).
-v {d0_v1,d1_v1}, --version {d0_v1,d1_v1}
                      Select the model to use.
-f FILE_PATH, --file FILE_PATH
                      Specify the input image for Inference mode instead of using camera. Ignored if evaluation mode is enabled.
-s, --save            Images with detection results will be saved (not when using camera).
-m MIN_SCORE, --minScore MIN_SCORE
                      Set the minimum score to display a detection (between 0 and 1 inclusive).
-nv, --noviz          Image(s) will not be displayed.
-c CAMERA_NAME, --camera CAMERA_NAME
                      Choosing camera/video stream to use. Default is 0.

Project Members

Name	GitHub profile
Adrien JAUGEY	https://github.com/AdrienJaugey
Bastien SUILLOT	https://github.com/Ricotimes