Medical image classification with fast.ai on Render

This repo is based in fast.ai model for Render.

The Deep Learning module has been trained with the HAM10000 2018 ISIC dermoscopic images, achieving 92,3% accuracy.

This application can be found in: https://skin.onrender.com. You can test it with your own images! It has been modified to support heroku.

You can test locally installing Docker and using the command:

docker build -t skin . && docker run --rm -it -p 5000:5000 skin

You can find the Render deployment guide in https://course.fast.ai/deployment_render.html.

Please use Render thread inside fast.ai forum for questions and support.