The LA Wildfires have left behind unimaginable destruction—displacing families, destroying homes, and wiping out wildlife habitats. With climate change fueling a global surge in wildfires, the threat is only growing. Wildfires are now responsible for nearly one-third of global CO₂ emissions. The key to minimizing their impact lies in early detection and rapid response.

That’s where Lume comes in—a smart, multi-point detection system designed to spot smoke and flames in real time and alert emergency responders before disaster escalates.

Lume is an AI-powered wildfire monitoring system that:

• ✅ Detects smoke and fire in images and live video feeds using deep learning.
• ✅ Processes video frame by frame in real time, marking fire zones with bounding boxes.
• ✅ Automatically alerts emergency services (911) when fire is detected.
• ✅ Features a user confirmation system—if the user dismisses the alert but the fire persists, the system continues prompting until resolved.
• ✅ Includes an interactive dashboard (Flask + JS + HTML + CSS) for real-time monitoring and alerts.

👉 Try it live here: Lume Deployment

• YOLOv8 → real-time fire detection
• OpenCV → video processing and frame analysis
• Flask → backend + API integrations
• JavaScript, HTML, CSS → dashboard UI
• Twilio API → automated emergency service alerts

• False Positives: Initially, YOLOv8 produced inaccurate detections. We improved performance by fine-tuning the model and applying data augmentation techniques (e.g., color transformations, contrast adjustment).
• Dependency Conflicts: Flask dependencies were incompatible with some Python versions—resolving this required debugging, testing, and version control adjustments.

• Integrating API services for automated emergency response triggered by sustained fire detection (≥5 seconds across multiple frames).
• The value of team collaboration—transitioning from solo projects to coordinated teamwork.
• Improving model accuracy through data preprocessing and augmentation.

• 🔹 CCTV integration for continuous wildfire surveillance
• 🔹 Google Maps API integration to pinpoint fire locations and share them with responders
• 🔹 A web-based detection tool allowing users to upload images/videos for instant fire threat analysis
• 🔹 Mobile app support for community-driven wildfire reporting