Inspiration was when I came to know that COVID-19 spread through aerosol particles indoors. Those aerosol particles do not settle on the ground for longer hours of time and keep circulating in the air. That is when I thought I should use Deep learning to prevent the spread of COVID-19
What it does
It detects cough/sneeze in real time deployed with a microphone. Thus alerting people or alternative methods can be used to prevent spread of COVID-19
How we built it
The microphone keeps recording live audio. When the audio detects sound above certain decibels it starts saving the audio for about five seconds of window size after which the audio is converted into a spectrogram and is sent into a neural network. The neural network differentiates cough/sneeze sounds from other environmental sounds.
Challenges we ran into
Challenges I ran into was when I thought my idea can be solved by computer vision. But deep learning with sound classification was a better solution to this problem. Another problem is optimizing the neural network, as the system had some latency in detecting the sounds.
Accomplishments that we're proud of
I am proud of trying to serve the humanity with technology. This can be an effective method of trying to stop COVID-19 spread indoors.
What we learned
We learned that AI can be used not only in computer vision but also in sound classification. AI does not discriminate between different types of data.
What's next for Live Indoor Cough/Sneeze detection
The next step would be to include wireless room freshners or other indoor spraying setups with the main system which can detect cough/sneeze. When a cough/sneeze is detected those sprayers can used to spray viral neutralizing friendly chemicals in air . The next step is also to reduce the latency involved in the system.