Land cover features and their changes are an important part of global knowledge and influence environmental management processes, political decisions, and planning under the auspices of international organizations. Mapping of land use/land cover changes is performed at several spatial and temporal resolutions, and during the past century, the process has experienced considerable steps forward, first, with the introduction of aerial photography and, second, with the development of satellites and remote sensing techniques (Senf and Seidl 2021; García-Álvarez et al. 2022; Song 2023; Bonannella et al. 2024). The latter has led to the availability of large quantities of data at various spatial and spectral resolutions, but the problem remains obtaining the desired information in a precise and automatic way to support continuous change detection and final product display in a digital map form. Currently, the accuracy of machine learning and deep learning algorithms in land use and land cover change detection has reached >80% (Sefrin et al. 2021; Yuh et al. 2023). At the global scale, the resolution of satellite images is generally between 3 and 30 m (for example, Landsat, Sentinel, and PlanetScope), and ready-to-use products are available through dynamic web map services (Hansen et al. 2013). With the rapid advancement of modern technology, other types of data are now available and can be used for land cover changes, especially deforestation and afforestation. They include data produced by laser scanning from satellites, air planes, and drones (unmanned aerial vehicles, UAVs). These data are provided as point clouds that provide information on laser beam signals reflected from ground surfaces, plants, buildings, etc. Points of known positions (x-coordinates, y-coordinates, and elevations) require classification to obtain correct information on the Earth’s surface and its features.