HYBRID CNN-RF AND CNN-XGBOOST FRAMEWORK FOR REAL-TIME FOREST FIRE DETECTION

The primary goal of this research is to detect forest fires as early as possible. It detects fires by analyzing camera images and movies. The work focuses mostly on picture recognition algorithms. Fire boundaries are constantly shifting, necessitating a series of steps beginning with background subtraction. The candidate regions are then identified using a color segmentation model. A Convolution Neural Network (CNN) is used to determine whether or not candidate areas have a fire. When an actual fire occurs, it rapidly labels it without using a sensor device and alerts the forest service.