RingCareAI

Never Unknow the Unsafe

Comment

Inspiration

Watching numerous OpenCV projects on social media , This Hackathon and the existing features , userbase of Ring Device inspired to start this app to make Ring devices more impactful for user security

What it does

RingCare AI connects to the Ring Playground and retrieves available Ring events and video footage. It then uses OpenCV to analyze video, generate a keyframe, and present the results alongside an event priority in a caregiver dashboard. Detects persons , weapons and more in ring device footage and notifies the user

How we built it

The pipeline: Ring event, then the FastAPI backend downloads the clip, OpenCV picks frames, Rekognition detects people and objects, the rules engine sets the priority, DynamoDB stores it, alerts go out, and the React dashboard shows it. Ring integration: the client in backend/app/ring.py, the token kept on the server, and the way clips are downloaded. Detection: OpenCV for motion and frames, Rekognition for people, weapon-like objects and face coverings, and a fallback to the local detector. Rules engine, no AI model: deterministic thresholds, the fall and security indicators, and the safety wording. Storage and alerts: DynamoDB with duplicate protection, SNS email, browser alerts and reminders. Testing and process: the automated tests, the staged clips, and the phase-by-phase plan.

Challenges we ran into

One of my biggest challenges was working with a simulated camera environment instead of physical Ring hardware. We needed to distinguish between successfully creating a live-view session and actually obtaining video that our own application could process. Since I could not fetch clips of robbery , attack and danger I installed them locally and then pushed them to the app through terminal [curl] so that I know how much accurate the app is !

Video playback and format compatibility presented another challenge. A stream that works inside the Playground is not automatically ready for playback in a browser or analysis by OpenCV

Accomplishments that we're proud of

The app immediately detects weapons like gun , knife etc. and flags the footage as danger . It also notifies the user for such instances and reminds them if the danger is not acknowledge making it useful for home owners , shop-owners and people who really want to know if anyone is approaching their property

What we learned

I learned about the ring playground and how to connect it to any app , Explored AWS Rekognition , DynamoDB , Handling videos and playing them in browser for better UX and also learned a lot about Kiro Code.

What's next for RingCare

If people genuinely like this app and it receives positive feedback from judges then I will add features to connect it to any ring device , deploy the app and make sure it fetches 24/7 without the user requiring to fetch manually and launch it as a mobile app too.

Built With

Share this project:

Updates

Submission history