PneumoDL

Hello everybody! Let me introduce you to our Project called PneumoDL.

We are a team of two students working on an artificial intelligence. This AI aims to detect if a person is ill from recordings of breathing sounds. To do this, we needed to train this AI and therefore we needed samples. Luckily, we found a large library of breath sounds available for free on the internet.

How are sounds processed? Each sound is first filtered and normalized . We then use python libraries to apply mathematical  transform to turn these sounds into images. These images, called spectrograms and scalograms, are then given to the learning algorithm. This one will learn to distinguish between a sick person from a healthy person.

BOUROUISS Yacine