The development of artificial intelligence systems allows the application of a set of technical vision algorithms to solve issues of deciphering high-spatial-resolution optical survey data obtained from various types of drones. This study aims to develop a technology for recognizing objects in logging sites. The study objects were cutting areas of varying-intensity logging. The visible-range survey data from a DJI Mavic 3 were investigated. The You Only Look Once version 8 (YOLOv8) model was applied as a computer vision algorithm, an advanced solution in the field of technical vision due to the high speed and accuracy of recognizing various objects in images. The neural network was trained on objects: growing trees, species composition, areas for storing logs, assortments, logging residues, and soil damage. Data labelling was implemented using the Label Studio software product, and the network was trained in the Python environment. The models available in YOLOv8 have five levels of image processing, which determine the accuracy of object detection and processing time. The level of processing is determined experimentally by the accuracy of object recognition for a specific task. All five models were tested to find the best solution for identifying various objects in the forest. The study results showed acceptable accuracy in identifying growing trees, classifying tree species composition, and determining storage locations for logging residues, stacks, and logs.