DETECTION AND RECOGNITION OF GROUND FIRES FROM AERIAL PHOTOGRAPHY IMAGES

The paper discusses the use of aerial photography to detect adverse and dangerous phenomena occurring on the Earth's surface. Such phenomena include various environmental problems, which include: environmental pollution, oil spills, debris accumulations, deforestation, fires, etc. The most dangerous problems are fires that can occur in a variety of places: in forests, fields, residential buildings, industrial facilities. Aerial photography is the most effective way to detect dangerous phenomena, in particular fires. Aerial photography is an important tool for obtaining up-to-date images. At the same time, facilities conducting aerial photography should be equipped with appropriate high-quality cameras. The resulting photos of the images must be recorded by GPS data throughout the filming to determine the coordinates of the terrain. The purpose of the work is to develop and implement an algorithm for recognizing fires in images obtained using aerial photographs. To solve the problem of fire detection and recognition in images obtained from aerial photography, the use of deep learning methods using neural networks is proposed. The article discusses the application of image recognition methods. The basic operations of convolutional networks are also considered. An algorithm for developing an image recognition model using a deep learning procedure based on convolutional networks is presented.