posted an update

https://youtu.be/GG1JZjmOChc

This is the video link we were not able to submit on time since it requires a youtube link.

Tech STack:

Frontend: We built the mobile app using React Native with Expo, which gave us easy access to native features like the camera, GPS, speech, and clipboard without writing native code.

Backend: Our API runs on Python with FastAPI, chosen for its speed and async support. We use Pydantic for data validation and Uvicorn as the server.

Database: All user profiles, face embeddings, and location history are stored in MongoDB Atlas, accessed through the Motor async driver.

Authentication: Firebase handles user registration and login. We generate custom tokens on the backend so patients can also log in using just their face.

Face Recognition: We use DeepFace with the Facenet512 model to extract face embeddings. When scanning, we compare faces using cosine similarity with a 0.75 threshold to identify family members.

Maps & Location: Mapbox powers our address autocomplete and live tracking maps. Expo Location handles GPS - patients send their location every 5 seconds, and family members see updates every 3 seconds.

Speech: When a family member is recognized, Expo Speech announces their name and relationship out loud to help the patient remember.

Github: https://github.com/billchenziang33/hackville1.17.git

Log in or sign up for Devpost to join the conversation.