DetectNow

Using DeepLearning to detect the cough caused by COVID-19 from sound recordings. Our vision is that that everybody can get screened at zero costs and directly from home.

This solution was built during a hackathon, checkout the devpost for complete details here.

Steps to Reproduce

Clone The repo

git clone https://github.com/shresthagrawal/covid19.git cd covid19

Create a virtual env

virtualenv venv

source venv/bin/activate

Clone

git clone https://github.com/tyiannak/pyAudioAnalysis.git

Install Dependencies

pip3 install -r ./App/requirements.txt

pip3 install -r ./pyAudioAnalysis/requirements.txt

brew install ffmpeg

Install

pip3 install -e pyAudioAnalysis

Using Test.py

python3 App/app.py

Alternative

chmod +x ./auto.sh

./auto.sh

Training on the original dataset

python3 pyAudioAnalysis/pyAudioAnalysis/audioAnalysis.py trainClassifier -i data/cough/not_sick data/cough/sick --method randomforest -o model_new

Training on Uploads

python3 pyAudioAnalysis/pyAudioAnalysis/audioAnalysis.py trainClassifier -i data/uploads/not_sick data/uploads/sick --method randomforest -o model/model_new

Training Dataset

data-set

DISCLAIMER:

The diagnostics function is not live yet, as this will require a medical trial first. By uploading your cough sound recording already now, you make a valuable contribution to getting this tool ready for a broader audience. Please take 10 seconds to fight COVID-19 and upload your sample here: https://www.detect-now.org/ (edited)