Inspiration
Inspired by real challenges from the emergency department, we designed a tool that helps doctors/paramedics identify unknown patients in seconds. Every moment matters in critical care, and our platform gives doctors and paramedics immediate access to the information they need to act fast.
What it does
Our tool allows clinicians to instantly identify patients by facial recognition. After snapping a picture, our program runs a comparison with the faces in the database. If a match is found based on consistency percentages, the patients profile is presented with medical information such as name, age, sex, height/weight, medical history, etc. Alongside this feature, we have an "Add Patient" field where you can enter a patient into the database, by filling out necessary information and taking a picture. Finally, the clinicians are able to browse the database, filled with all of the patients and their medical information.
How we built it
Backend- Built with FastAPI, DeepFace, Open CV and numPy. Open CV and DeepFace work with the webcam to power our facial recognition feature. The facial recognition is then compared with images in the database, returning a match if found. Frontend- Using react, we built a clean and uncluttered UI that doctors can use to navigate the features.
Challenges we ran into
The "Add a patient" section gave us a lot of trouble, updating the database took up most of our time. Another issue we faced was accuracy of the facial scan. The solution we came up with was two simultaneous photos that both enter into the database, giving the algorithm a better chance of successfully matching the photo with a patient.
Accomplishments that we're proud of
We are proud of the level of consistency our algorithm was able to produce in such a short amount of time. Except for a couple hiccups, we managed to get it going pretty consistently.
What we learned
Aside from technical skills, we learned how to take all of our skills together to make the best of our time. Each person had a task that we worked on and it made all sides of our product good enough to be presented/demoed.
What's next for IdentiCare
To actually implement this system, we need a stronger database to start, along with a proper server to properly be implemented in hospitals. Next steps for us is to spend more time perfecting what we already have established, before expanding our horizons.
Log in or sign up for Devpost to join the conversation.