ResQ.ai

Inspiration

Many seniors and people living alone experience falls that go unnoticed for long periods. We came across stories of individuals lying on the floor for 30–60 minutes before help arrived. Wearables often failed because people forgot to wear or charge them, and panic buttons required the person to be conscious and able to reach them. We wanted a solution that never depends on memory or effort-a system that is automatic, reliable, and always watching out for loved ones.

What it does

ResQ.ai continuously monitors a person using a real-time camera feed and pose estimation. When a fall is detected: A 30-second alert is sent to the user’s phone. If not cancelled, the system notifies emergency contacts. If still unanswered, it escalates to an automated call. No wearables. No buttons. Fully hands-free safety.

How we built it

Computer Vision: YOLO-Pose is used for real-time posture detection. We analyze centroid drop, velocity, and angle shifts to detect falls. Backend (FastAPI): Handles fall events, user profiles, and escalation flow. Mobile App (Flutter): Displays alerts, provides the cancel button, and handles notification logic. Database (MongoDB Atlas): Stores users, incidents, contacts, and history. Pipeline Flow: Camera → YOLO Pose → Fall Logic → FastAPI → Flutter App → Emergency Contacts.

Challenges we ran into

Integrating Python backend with Flutter frontend Ensuring YOLO runs smoothly in real time. Rewriting backend schemas when switching from SQL to MongoDB. Implementing reliable 30-second alert and escalation logic. Debugging Flutter + FastAPI + CV pipeline across multiple devices.

Accomplishments that we're proud of

A working fall-detection pipeline with real-time inference. Clean integration from vision model to backend to mobile app. Low latency detection with stable performance. A usable solution with real-world impact potential. A polished UI and smooth user experience. Integrated SMS + voice calls with Twilio.

What we learned

Vision-based fall detection using pose keypoints and motion metrics. Full-stack integration using Flutter, FastAPI, and MongoDB. Designing safety-critical notification flows. Effective teamwork and fast iteration during a hackathon.

What's next for ResQ.ai

Deploy on edge devices (Jetson Nano / Raspberry Pi + Coral). Train a custom dataset for improved fall accuracy. Build a caregiver dashboard for monitoring trends. Expand to detect inactivity, wandering, or health anomalies.