What it does
Currently it classifies X-rays as Normal or Infected (Pneumonia) with 92% accuracy. Pneumonia/Normal dataset was used because the dataset was readily available online and the Pneumonia X-rays are very similar to covid-19 X-rays.
How I built it
Because the data was very limited, the X-ray images were first standardized and then augmented to produce as many copies (variant copies) as possible. This way the network could be trained far better. Speaking of the network, a convolutional neural network was implemented to pick up on key features and classify the X-rays.
Challenges I ran into
Arranging the data into it's labels and image contents
Creating batches of data
Accomplishments that I'm proud of
Successfully training the model and achieving 92% accuracy.
What I learned
Convolutional neural networks
What's next for Pneumonia(Covid)-Detector.io
Create a similar model but exactly for Covid-19 once the data becomes available.