Welcome to Drone-Powered SAR, a project that combines the power of AI 🤖, drones 🚁, and thermal imaging 🔥 to save lives in critical search and rescue missions. This project leverages advanced technologies to redefine how we locate missing persons in challenging terrains like mountains 🌄 and forests 🌲.
- Multi-Mode Imaging: RGB 🌈, thermal 🔥, and composite overlays to ensure accurate person detection in any environment.
- YOLOv4 Implementation: Lightning-fast image classification for real-time victim identification 📸👣.
- Optimized Pathfinding Algorithms: Quadrant Scanline Fill 🗚️ ensures the fastest drone flight path to locate victims.
- Unity Simulation: Realistic SAR scenario testing in a 3D environment 🌐, modeled after Whistler, British Columbia.
SARSimulation3.mov
- AI-Enhanced Detection: Train YOLOv4 models with RGB, thermal, and composite datasets for superior detection accuracy 📊.
- Efficient Pathfinding: Compare and implement the best drone navigation algorithms 🫅 for rapid victim recovery.
- Real-World Impact: Equip drones with dual-mode cameras to improve SAR operations in diverse conditions 🌥️🌇.
- Drone Imaging: Remotely Piloted Aircraft Systems (RPAS) for autonomous or semi-autonomous operations.
- Neural Networks: YOLOv4 for real-time image recognition and object detection 🤖.
- Unity: Simulations for SAR scenario testing in dynamic terrains 🎮.
- Python: Data preprocessing, annotation, and composite image generation 🐍.
- Precision: Achieved 98.65% accuracy with composite imaging ✨.
- Time Efficiency: Quadrant Scanline Fill reduced search times by 36.76% ⏱️.
- Real-World Feasibility: Composite datasets proved effective in diverse lighting and weather conditions.
- Create custom drone datasets for even more realistic training scenarios.
- Optimize Unity simulations for enhanced SAR operations 🛠️.
- Implement live video relays for real-time processing 📹.
- Test multi-drone coordination for faster, collaborative missions 🤝.
For detailed insights into the research and results, check out the attached paper 📜!
We welcome feedback, ideas, and collaborators to help expand this project’s life-saving potential. ✨
Vancouver - Summer 2021 - Northeastern University