Bandaid.AI

Front-End, Home page
Front-End, Image Submission

Inspiration

One of our friends got bitten by a dog a few days ago, and as we were helping them clean the injured area, we had no idea about what to do next. None of us had been bitten by dogs before, so none of us had any previous experience in dealing with this kind of emergencies. An idea popped up in our mind: what should students do in this sort of emergent situations, where they are injured, anxious, and don't know whether they should seek medical support? We thus decided to create an AI tool that utilizes vision and language models to help students get a better sense of their injuries.

What it does

Our program has two main functions: injury severity classification and medical advice. Students can take a picture of their injured area, upload it to the website, and get a predicted severity score indicating how severe their wound is. Then, they can describe their injury details to the website, which generates a thorough medical advice based on user input.

How we built it

Start both ways from frontend and backend. The front end build a platform that handle user login, user profile entering, and the metadata would be saved into a JSON file that being passed to backend. The backend will grab using data in a parallel setting. We will put the user injury image to our self-trained AI to get a wounded level output; The other one is for a text paragraph input, which takes the user description and send it to AI, which we used API key to achieved. Finally the AI will output suggestions that will be send back to the user. The deployment of AI model to our backend code is hard since there are multiple different dependencies for us to add on and we need to make sure they are all compatible.