Realtime Mask Detection

Machine Learning Project purely meant for educational purpose

Problem statement

In today's pandemic world, it is crucial to monitor the people in the world. It is a necessity to wear mask to prevent the spread of contagious communicable diseases that spread through the air, especially CoVID-19. A need to maintain personal hygiene to avoid spread of disease has motivated people to wear masks. But, not all people follow the norm. So, the application at hand monitors the people at realtime to detect if a person wears a mask or not.

Install Dependencies

If you have already installed anaconda, you can create a virtual environment using the following commands:

conda create --name env
conda activate env

Make sure to install the python packages in requirements.txt:

pip install -r requirements.txt

Dataset

• Sourced from:
  • Kaggle
  • Google Images
  • Other open source image libraries
• Credits: Balaji Srinivasan

Repository Structure

• dataset/ contains the dataset for training the model
  • with_mask\ contains images of people wearing mask
  • without_mask\ contains images of people NOT wearing mask
• face_detector/ contains the model that detects the face, which the region of interest to detect the mask
  • deploy.prototxt is the protobuffer text file that contains details regarding the face detection model
  • res10_300x300_ssd_iter_140000.caffemodel is the model used to detect faces
• Train.py: contains the code to preprocess the images in the dataset and train the model to detect masks
• Test.py: contains the code to start a video stream to perform a realtime mask detection on the live stream

References

1. Face Mask Detection by Balaji Srinivasan - YouTube Tutorial and Walkthrough of code
2. Face-Mask-Detection by Balaji Srinivasan - Github Repository
3. Tensorflow installation (with CUDA, cudNN and GPU) by Aladdin Perrson - YouTube Tutorial