Skinmergency

Our title page
Main screen
Information
Description
Treatment
WebMD Link
Image Comparison

Inspiration

The inspiration for our application came from our own laziness and unwillingness to go to the doctors for anything. We wanted to create an application with machine learning and artificial intelligence models in order to improve the daily lives of the lazy. Our goal is to create an application to change the future of skin care and self analysis of skin conditions at home.

What it does

Our application takes in a picture of the user's skin taken using their phone camera and uses that picture to conduct a smart inference on what skin condition the user might have based on our trained machine learning model that uses image classification.

How we built it

For our front end, we used android and java to build the ultimate experience for the mobile user. We also used android's built in camera in order to snap a picture of their skin. For the machine learning model, we used Microsoft Azure cloud computing platform. More specifically, the custom vision AI.

Challenges we ran into

One challenge we ran into was figuring out how to use TensorFlow to integrate the trained model, the camera and android studio. Additionally, the training of the model was challenging because the platform that was used is not extremely sensitive to subtle differences. The data set that we used had to be carefully selected in order to increase the precision and accuracy of our model.

Accomplishments that we're proud of

We, as a team, are proud of the fact that we reached our front end goal and finished a complete, marketable application. We are also proud of our machine learning model and android integration. Our final user interface is user friendly to its fullest extent.

What we learned

We, as a team, learned how integrate a Microsoft custom vision AI and android. We also learned how to create a sleek android user interface for the user. Some team members learned how to step aside and ask other team members for their support and input to their part of the project. This got the foundation of the application running and boosted team morale.

What's next for Skinmergency

We would like to increase the accuracy of our current model and we would also like to start including and supporting other well known skin conditions. The end goal of Skinmergency is to get real doctor certification in order to improve trust, and reliability of our machine learning model. We would like to possibly take our application to enterprise and become a trusted source of skin condition diagnoses.