We wanted to build something good after our first idea failed

What it does

Given a picture of a "mole" it classifies it.

How we built it

We scrapped data from the web and used HAM10000 dataset from kaggle to build an autoML to detect "moles". We use a flask server to interact and manage the states.

Challenges we ran into

Time, had less than eight hours to build it. High network traffic, had to compromise between datasets. GCP documentation.

Accomplishments that we're proud of

Making something useful in such a short time.

What we learned

cleaning data, auto ml and GCP.

What's next for Dignaosaurious

we would like to better train the skin model and then add other easy to diagnosable diseases like heat beat murmmers and more.

Share this project: