Inspiration
What it does
How we built it
Challenges we ran into
Accomplishments that we're proud of
What we learned
What's next for FireSentinel-AI
Inspiration
Smart TVs in living rooms are usually used only for entertainment, while household security alerts are limited to mobile phones. We realized that critical safety warnings (like kitchen smoke, gas leak, or unattended flames) often go unnoticed while watching TV. FireSentinel-AI was built to transform Smart TVs into active ambient safety guardians for Amazon Fire TV (Vega OS).
What it does
FireSentinel-AI works in the background by analyzing real-time camera streams using OpenCV 5. When a fire hazard, thermal anomaly, or smoke pattern is detected, the AI agent makes an autonomous decision to display an instant, non-intrusive Picture-in-Picture (PIP) Overlay Alert on the TV screen without interrupting the user's entertainment.
How we built it
- Vision Engine: Built using Python and OpenCV 5 to process video frames and detect environmental hazards.
- Frontend & Overlay: Custom web interface designed for Amazon Vega OS compatibility to display PIP alerts.
- Data & Logging: Integrated with Supabase to store event logs and alert history securely.
- Repository: Hosted on GitHub with a complete open-source workflow and custom MIT license.
Challenges we faced
- Optimizing frame-by-frame computer vision processing on TV hardware without impacting streaming performance.
- Designing a privacy-first architecture where all video processing happens locally at the edge.
What we learned
- How to leverage OpenCV 5 for multi-step agentic perception-decision-action loops.
- Developing privacy-preserving vision workflows for TV platforms like Vega OS.
Log in or sign up for Devpost to join the conversation.