/Cough-Detector

Pnuemosense: A challenge to build a web based app for detecting cough sounds on live video and perform risk analysis

Primary LanguageCSS

Problem Statement

To monitor the presence of face masks and detect coughs in real time and classify individuals into varying risk categories.

Built With

  • Flask
  • Flask-SocketIO
  • Librosa
  • Keras
  • Tensorflow
  • OpenCV
  • Media Recorder API
  • Bootstrap 4
The Web Application is deployed on Microsoft Azure, and can be accessed via

https://pnuemosenseai.azurewebsites.net/ [Unavailable now]

Troubleshooting

  • It is advised to open the link in the incognito mode.
  • Latency: latency depends on the GPU of the system on which you are running the web application. For best results, A high performing GPU is required.

How do I deploy the app?

Getting Started

The following package versions must be installed to successfully deploy the model.

  • matplotlib 3.1.3
  • ffmpeg 3.2
  • numpy 1.18.5
  • Flask 1.1.2
  • gevent 1.4.0
  • Keras 2.3.0
  • librosa 0.6.3
  • numba 0.49.1
  • tensorflow 2.1.0
  • python 3.7.1
  • flask-socketio 4.3.1
  • imutils 0.5.3
  • opencv-python 4.3.0.36
  • six 1.12.0
  • scipy 1.4.1
  • setuptools 41.0.0

To run the model on your local machine,download this repository as a zip file, clone or pull it by using the command

$ git pull https://github.com/mitali3112/Cough-Detector.git

or

$ git clone https://github.com/mitali3112/Cough-Detector.git

Requirements can be installed using the command (from the command-line) preferably in a virtual environment.

$ pip install -r requirements.txt

Then, navigate to the project folder and execute the command

$ python app.py

to deploy the app locally.

On your web browser go to http://localhost:8000/

Demo

![COVID-19 Risk Assessment App Demo](cough_detector_demo (1).mp4)

Contributers

  • Aparna Ranjith
  • Gunveen Batra
  • Mansi Parashar
  • Mitali Sheth
  • Sruti Dammalapati

Acknowledgements

We thank B-Aegis Life Sciences for the opportunity.