Application of artificial intelligence neural networks in photogrammetric processing of digital data

The article examines the impact of artificial intelligence neural network algorithms on the process of photogrammetric processing of digital images and the formation of dense point clouds. It is noted that the integration of machine learning algorithms, in particular deep learning, allows you to automate key processing stages, increase the accuracy of object classification, optimize geometric correction of images and improve the quality of final geospatial products in the form of an orthophotomap of the area. The main stages of the photogrammetric process are described: from collecting primary data to forming orthophotomaps and three-dimensional terrain models. Particular attention is paid to the role of the neural network in improving the density of the point cloud due to the correct interpretation of points on digital images, object classification and reducing the influence of the human factor. The challenges associated with the need for large volumes of high-quality training data are identified. The prospects for the application of AI neural networks in photogrammetry and the need for further research in this direction are substantiated.