DermaML

Results after uploading
Upload Page
Home Page

Inspiration

When I was young I had a lot of moles which appeared to be dangerous, had to get biopsies, it was very intrusive and scared me a lot, I really wished that there was a less intrusive way to detect them without scaring me as much so I created this

What it does

Uses a camera to detect moles, then utilizes an advanced machine learning algorithm trained on HAM10000 to detect similar skin diseases

How we built it

We used React/NodeJS as our Frontend Python with FastAPI as our backend TigerData as our DB

Challenges we ran into

We struggled with the CORS system not syncing, which caused issues with the data not displaying. Setting up a local API server; we tried Render, but it ran out of space. Tried hosting on different sites; they cost money, till we ended up using ngrok to self-host from the laptop. General problems with how React worked caused complications with our front end

Accomplishments that we're proud of

Connecting the API server to the Front End accurately Creating a Machine Learning Algorithm within 2 hours while maintaining a high accuracy

What we learned

We learned how to connect the frontend and backend using api servers, how to set up local hosting for our api, and how to train an MLA to detect possibly dangerous moles

What's next for DermaML

If we had more funds, we ideally would have implemented a hardware aspect to this project, as we really wanted to create a physical scanner device which would allow for more accuracy using sub-surface light scattering; however, the cost is out of our budget.