Catch95

This is a Web-Application which uses a CNN-Model to detect in real-time if a person is wearing his mask properly or not and if the type of mask he's wearing is N95 or not. This has wide range of applications- mostly for monitoring and gauging the risk levels at public places and raising an alarm to inform the authorities.

How to get started with the Code:

The following dependencies must be imported before running the code.

• pip install opencv-python
• pip install imutils
• pip install keras
• pip install tensorflow
• pip install django
• Make sure that your device has access to a web-cam for the program to run.

After installing the above listed dependencies :

• Download the zip-file and extract the contents.
• There are a few paths that need to be changed on your system locally:
  • go to Catch95
    • firstproj
    • app1
    • hope.py - on line numbers 24, 27, 32 change the paths to the paths of your locally saved haarcascade_frontalface_default.xml, model-best and app1/keras_model.h5 respectively.
• Add python and pip to your system path(if they are not already). Refer to this website if facing problems: https://datatofish.com/add-python-to-windows-path/
• Open the anaconda powershell or cmd terminal if using Windows, navigate to the directory containing the filename "firstproj" and run python manage.py runserver to set-up the Django server.
• Open up Google Chrome and type in localhost:8000\ or your IP:8000. Catch95 is up and running!

OR

• Clone the github repo on to your computer typing !git clone https://github.com/ASHWarriors/Catch95.git on the Git Bash Terminal.
• There are a few paths that need to be changed on your system locally:
  • go to Catch95
    • firstproj
    • app1 - hope.py on line numbers 24, 27, 32 change the paths to the paths of your locally saved haarcascade_frontalface_default.xml, model-best and app1/keras_model.h5 respectively.
• Add python and pip to your system path(if they are not already). Refer to this website if facing problems: https://datatofish.com/add-python-to-windows-path/
• Open the anaconda powershell or cmd terminal if using Windows, navigate to the directory containing filename "firstproj" and run python manage.py runserver to set-up the Django server.
• Open up Google Chrome and type in localhost:8000\ or your IP:8000. Catch95 is up and running!

Model Accuracy :

To know the model accuracy

• open Catch-95.ipynb using Jupyter Notebook and inspect the accuracy.

Model	CNN Model	Google Teachable Machines
Accuracy	84.38%	uptill 100%

DataSets acquired from:

• https://google.com
• https://www.kaggle.com/omkargurav/face-mask-dataset
• Self taken photos

Resources Used:

• https://getbootstrap.com
• https://w3schools.com
• https://www.free-css.com/free-css-templates
• https://teachablemachine.withgoogle.com
• https://www.pyimagesearch.com/2017/09/18/real-time-object-detection-with-deep-learning-and-opencv/
• https://github.com/achen353/Face-Mask-Detector
• https://stackoverflow.com

A Demo of the Project:

https://youtu.be/QxJjTHvaIgc