Dermi focuses on helping the community by having early diagnosis of skin diseases so that treatment can happen sooner, reducing deaths, and it is less expensive (reducing the burden on systems such as Medicare).
Inspiration
Many die and become disabled every year from skin diseases. If early detection can solve the issue, we need to make it more available. I aim to do that using deep learning.
What it does
Suggests skin disease from image.
How I built it
First I collected a dataset using a web scraper. Then I created a ResNet50 model using Pytorch.. Afterwards, I created a Flask server for inference and used Ngrok to make it publicly available. Created an Android app using Kotlin and Room for on-device database.
Note: long press on a diagnosis allows you to delete it.
Accomplishments that I'm proud of
I am proud of the application and the speed at which I built it.
What I learned
Learning how to create a ngrok server on Colab itself.
What's next for Dermi
Larger dataset Using several models to be able to create the most accurate diagnosis. The current app uses 1 model, however, I previously created another model with another dataset, which I chose not to use for this hackathon due to the time restriction. Using other available datasets (this app can integrate with other models, making it extensible) iOS application
Log in or sign up for Devpost to join the conversation.