/face_mask_detector

Face Mask Detection and Tracking with OpenCV and TensorFlow

Primary LanguagePythonMIT LicenseMIT

Introduction

The Year 2020 is dominated by the global COVID-19 Pandemic. People wearing Face Masks, to protect themselves against an Infection, is an omnipresent Picture.

This Project uses a Combination of OpenCV Face Tracking and a TensorFlow Neural Network to detect Faces and distinguish between People wearing a Face Mask and People who don't.

Those Informations are used to control an automated Camera-Mount, to track People not wearing a Face Mask.

Project Structure

1. Face Tracking

  • OpenCV is used to capture a live Video from a Webcam
  • Using the Deep Neural Network capabilities of OpenCV with a pretrained Caffe Model to implement Face Tracking

2. Face Mask Detection

  • Training a custom Convolutional Neural Network with MobileNet as a base Model

  • The tracked Faces from the Webcam Feed are used as an Input for the CNN

  • Based on the prediction the Box Color and Label changes (green for Mask and red for No Mask)

3. Automated Camera-Mount

  • Using the XY-Coordinates of the Face Tracking, an Arduino controlled automated Camera-Mount follows detected Faces without a Face Mask

General Usage

  1. Clone the Git repository: git clone https://github.com/samuelrothen/face_mask_detector.git
  2. Install the requirements: pip install requirements.txt
  3. Run /src/live_video_detection.py to start the live Video Detection

Arduino Usage

The usage of an Arduino is disabled by default. If you want to use an Arduino to control the Camera Positioning, follow these Steps:

  1. Connect a Servo Motor to your Arduino using the following Wiring:

  1. Upload the .ino-Sketch from src/aduino_sketch/python_servo_sketch/python_servo_sketch.ino using the Arduino IDE

  2. Open /src/live_video_detection.py and set use_arduino to True (default is False) and define your Arduino COM-Port in serial.Serial(...)

use_arduino = True
if use_arduino:
    arduino = serial.Serial('COM3', 9600)
  1. Run the Python File

License

Distributed under the MIT License. See LICENSE for more Information.