DermaScan

• 

• 

• 

• 

  Training Validation Loss

• 

  Training Validation Accuracy

Inspiration

Over 85 million people in the U.S. are affected by skin diseases. 50% of Americans over the age of 65, have more than one skin disease. To address this problem, we have designed a skin disease detection AI app that will allow you to scan your skin whenever you want to stay healthy.

What it does

A user can upload an image of the area of their skin possibly affected by a skin disease, and our app will send a prediction of what disease it is based on our data-driven AI model.

How we built it

For the front end, we utilized the React framework with React Material UI and Axios to handle HTTP. For the back end, we used ExpressJS for routing and HTTP, multer for HTTP request parsing, and TensorflowJS to interpret the model. We used Tensorflow to build said model.

Challenges we ran into

Sending files between the client and server was a large hurdle to overcome because the client was sending multiple requests instead of one at first and the file object being sent was read with the wrong encoding. Additionally, we had issues reading the TensorflowJS model.

Accomplishments that we're proud of

We were able to send files to and from the server. We implemented parallax scrolling and typewriter effects to give the UI a crisp, modern feel.

What we learned

We learned how to use ExpressJS, Axios, and Tensorflow.

What's next for DermaScan

We would like to find a larger and better dataset for a more accurate model. We would also like to expand this into giving advice based off of the condition detected.

Built With

• axios
• css
• express.js
• git
• github
• html
• javascript
• node.js
• python
• react
• tensorflow
• tensorflowjs