Real-Time Peatland Fire Detection Using Infrared Cameras and SVM Algorithm

This paper presents the development of an early detection system for peatland fires utilizing a quadcopter equipped with an infrared camera. The autonomous operation system is capable of recording flight data and transmitting it in real time to a ground control station. The detection mechanism employs the Support Vector Machine (SVM) algorithm, known for its robustness in classifying fire hotspots effectively. Comprehensive experiments were conducted by simulating various peatland fire scenarios to rigorously evaluate the performance of the system. Real-time testing revealed that the detection accuracy reached 97.56% from an altitude of approximately 5 meters when identifying hot spots. Furthermore, experiments assessed the limits of detection resolution, demonstrating the system's ability to identify hotspots with a diameter as small as 4 centimeters from a height of 9 meters. This research significantly enhances the capabilities of fire surveillance in peatland areas, providing a reliable tool for environmental protection and effective fire management. The findings indicate that integrating infrared imaging with advanced machine learning techniques can lead to improved response strategies in mitigating peatland fires.