Covid-19 has taken many lives from the world. Hundreds of thousands of lives have been lost in the US alone, proving to be a problem physically and mentally. In fact, hospitals are now receiving too many patients to handle, leading to a lack of care for many patients. As a result of this epidemic, my teammate and I have decided to create a machine learning model that can take COVID-19 chest X-RAYS, and return a percentage probability of the patient with that X-RAY having COVID. This has a great amount of applications, such as automation so that nurses do not have to manually scrape hundreds of X-RAY files at a time.
What it does
This is program utilizes a Machine learning model our group has trained in order to detect the probability of COVID present in an X-RAY photo
How I built it
We built this utilizing Python, Keras, and Tensorflow, creating a CNN that takes images, stores it as an array, and then transforms it into 4 dimensions, such that the model can take this input and return a probability based on how it was previously trained
Challenges I ran into
We had issues preprocessing our data such that inputted images would work with the classifier. We also had issues with training our model, as it would only run through one training generation, which was a huge problem
Accomplishments that I'm proud of
We are proud of the great accuracy that our model provides. It has near 90 percent accuracy, which is amazing considering how important this implementation is
What I learned
We learned a lot about how to preprocess images such that they fit in CNNs and are able to be classified.
What's next for COVID-19 Detector in X-RAY Images
The next step would be deploying this model in a webapp, which we tried to do but could not due to cross origin domain restrictions