Covid19 Detection App with Deep Learning

CovidAI App Prototype

Inspiration

What inspired us was the recent Omicron situation and how in general Covid19 has affected our life. We also decided to make a prototype since we were inspired by Figma that we got introduced a bit before the hackathon.

What it does

We first used MindSpore to train our custom data of Covid19 CT scans with the GoogLeNet convolutional neural network. On the second hand, we made a prototype of an app that would use the implementation of the trained model. Though we did not manage to implement it, but our prototype will simulate what we essentially would want it to look like.

How we built it

We build our model and trained it using MindSpore in the Huawei AI Platform. The data we used were from open public datasets. For the prototype, we used Figma.

Challenges we ran into

Mindspore framework was really different to other frameworks we are used to such as Keras. Getting used to know how MindSpore works and how we can train our model was challenging.

Accomplishments that we're proud of

We are proud that we managed to train our model with MindSpore and that we designed our first ever prototype on Figma.

What we learned

We learned a lot from this project. We first learned how MindSpore works and how we can create datasets and train the models. We also learned how to design an app with Figma. Overall, the whole process of developing and using machine learning was a good learning experience.

What's next for Covid19 Detection App with Deep Learning

To improve our model, we could use transfer learning with pretrained models. We could also use image enhancement techniques and super resolution to improve image quality. Another point would be to use more data.