Forest Fire Detection in Satellite Images Using Conventional Neural Network

The number of forest fires is increasing significantly. According to the NFS, 75% of forest fires are human-caused and the remaining 25% are caused by natural phenomena. The reason for the growing number of forest fires is hunting, which interferes with forests, especially in underdeveloped countries. A forest fire has a significant effect on the atmosphere and wildlife. This study used Landsat satellite images on the Google Earth Engine, to present a model that can identify the presence of fire in an image data. The data used here are images from Google Earth Engine. We use a machine learning model called deep Convolutional Neural Network to solve the forest fire problem. CNN is a supervised type of machine learning, most advantageously used in image recognition and computer vision. Using the dataset, the features of both types of images, i.e., fire and non-fire images, will be extracted. And then, based on the training data set, a forest fire can be detected.