In recent years, the National Institute of Aeronautics and Space (LAPAN) has used hotspot data derived from satellite imagery to identify and detect forest and land fires at an early stage. Hotspot data has greatly facilitated firefighting operations and enhanced enforcement activities. Nevertheless, the system has certain limitations, mostly stemming from its incapacity to differentiate between forest and land fires and other sources of heat or fires produced by typical human actions. In addition, this approach requires time-consuming verification and significantly depends on human elements for sophisticated analysis and validation. Lately, the field of deep learning has been implementing a novel strategy by making progress in the field of artificial intelligence. The algorithm has been trained to identify burnt areas by analyzing satellite images recorded between 2017 and 2019. It recognizes the pattern and tone of the image in these areas. To validate the presence of burnt areas, it compares the current imagery from the past week with the historical Sentinel-2 imagery for each cluster, specifically for forest and land fire identification. Initially, the satellite images are obtained and the noise is eliminated using a median Butterworth filter. Next, the characteristics of the area of interest may be grouped using a K-density-based agglomerative method. The hotspot may now be accurately detected utilizing the advanced deep memory Alex neural network. The outcomes of the hotspot identification procedure, which has an accuracy rate of 99.7%, may aid firefighters in promptly extinguishing flames and help law enforcement authorities in identifying the optimal target area. Hence, the recommended technology has the potential to enhance the efficacy and productivity of resources assigned by law enforcement agents, resulting in improved and more prompt public services.
