MeLanie | Devpost

Overview of user interface for web application.

Inspiration

Our team was fascinated by the application of image recognition in the health care industry. Ideating through a potential app that can detect abnormalities in MRI scans or X-rays but were hesitant as patients wouldn't have much use for the model as they wouldn't be taking X-rays at home. We ended up coming up with the idea to test for skin cancer so that patients and doctors would both benefit.

What it does

MeLanie is a Web App that uses computer vision to detect melanoma from lesion images.

How we built it

Flask was used to build the web application and Keras for building the CNN model for predicting melanoma. The entire project was built in Python.

Challenges we ran into

Configuring environment variables was a reoccurring issue for us requiring us to set up and reinstall applications several times. When we tried to test the model on real-world images the model did not have the expected accuracy.

Accomplishments that we're proud of

A big accomplishment for us was being able to integrate all the separate parts of the project. Being able to take the ML model and deploy it on the image upload page. Additionally, being able to train the ML model was also a great feat.

What we learned

Our group members' experiences varied a lot but we all were able to take away a lot of new information. Such as learning various python commands in terminals, Github, ML models, frameworks and project development, website development, debugging strategies, and coding collaboratively.

What's next for MeLanie

MeLanie could be improved in its accuracy as a short term goal, but long term MeLanie could be used to analyze and interpret more than just skin cancer and could be expanded to interpreting X-rays, MRIs and even more medical imaging data.