Deployment of a UAV-Based Fire Detection System

In this paper, we design and implement a fire detection algorithm using an unmanned aerial vehicle (UAV). In particular, we consider a scenario where a UAV is employed to find a fire spot where the fire is in its early stages. To this end, we first propose a path planning algorithm where the goal is to make sure that the UAV finds the fire in the shortest amount of time. During this stage, the UAV visits different parts of the area, takes an image from each part, and sends the image to the control center for further image processing. Then, we deploy a machine learning (ML) model, which is a residual neural network (ResNet), to process the image and determine whether a fire is detected or not. The ML algorithm has been trained using images taken by the drone and images from the Internet. Through experimental results, we show that the proposed path-planning algorithm along with the ML model can detect fire efficiently. The problem and the proposed solution in this paper can be also applied to a search and rescue scenario, where for example, a hiker is missing in a remote area.