Melanie - The Pocket Dermatologist

An iOS application that diagnoses skin cancer by taking pictures of your skin.

Inspiration

We're both mole-y dudes. After becoming concerned about a potentially cancerous mole, we decided to write a neural network that can look at a mole and diagnose whether or not it's cancerous, as well as an iOS application to interface with it.

What It Does

1. Open the application
2. Take a picture of the area in question
3. Receive your results and confidence levels

How We Built It

We're early in the development phase, but the current plan is for a phone to take a picture, resize it, and send it to AWS Lambda. Then, AWS Lambda should run the picture through a neural network hosted on an AWS EC2 instance, and returns the diagnosis and confidence

Challenges we ran into

Still unsure how to set upa web server that our application can send images to. Auto-layout is also so, so frustrating

Accomplishments that we're proud of

Learning tensorflow was a hassle, but very rewarding!

What we learned

We basically learned everything from scratch

What's next for Melanie

Identifying more skin conditions

Built With

• python
• swift
• tensorflow