Fire detection system based on unmanned aerial vehicle

Because of the low cost, strong mobility, and wide aerial view, the UAV is more and more widely used in the field of inspection and emergency rescue. Most of the traditional fire detection methods are based on the RGB color model, and their detection speed and accuracy are inadequate. In this paper, a fire detection method based on an autonomous drone platform is proposed. The drone flies on a designated route and carries an Ultra96-V2 development board with YOLOv3 fire detection algorithms deployed, which acts as an edge computing device to transmit the detection results back to the ground station in real time. Experimental results show that the recognition rate of the algorithm is 80%, the model memory compression is more than 75%, and the real-time detection frame rate is more than 3 FPS.