As part of the "Artificial Intelligence Technologies in Real Estate Registration" training program at the Digital Department of Moscow State University of Geodesy and Cartography (MIIGAiK), students are interning at KB "Panorama" JSC. During the internship, participants learn about the capabilities of GIS "Panorama" and the "Panorama Vision" Automatic Recognition and Vectorization Complex.

One of the key focuses of the internship is studying technologies for automatic recognition of real estate objects from satellite images using neural networks. This is the most promising method for automatic addition and updating of cadastral information. Neural networks are trained based on manually recognized land plots, buildings, and structures. This method isn't perfect and can lead to inaccuracies since source images of each area are unique. Buildings across the country's regions can be constructed differently, resulting in neural networks trained on one type of terrain performing poorly when recognizing objects in different terrain types.

The interns' practical work involves preparing training datasets for the neural network models of the "Panorama Vision" complex. Tasks are performed in the professional GIS "Panorama" environment, for which temporary licenses were issued. During the work, students receive consultations from highly qualified company specialists. Interns solve practical problems of establishing buildings' true positions considering source information issues and construction conditions. Students received initial instruction, assignments, and methodological materials reflecting the specifics of digitizing training sample objects, considering private sectors, urban development, metal hangars, collective garages, and other objects. The results will be used for subsequent automation and improvement of cadastral activities in identifying objects not registered in the cadastre. The "Panorama Vision" complex is designed for automatic recognition and vectorization of agricultural lands, hydrographic features, and capital construction objects using Earth remote sensing data (satellite and aerial photographs).

"Digital Departments" is a project implementing professional retraining programs. These programs are part of the federal project "Development of IT Industry Human Resources", which is part of the national program "Digital Economy of the Russian Federation." This joint project between the Ministry of Education and Science and the Ministry of Digital Development of Russia allows students to obtain a new profession in information technology.

KB "Panorama" and MIIGAiK are long-standing strategic partners. Their joint activities are aimed at promoting domestic cartographic and geodetic education and geographic information systems software.

KB "Panorama" has developed GIS ToolKit Active version 15.0.6 for creating geoinformation systems with free licensing in any programming environment supporting ActiveX technology, such as Visual Studio. The new version features an updated GIS core with enhanced RSC digital classifier management tools and overlay operations. Support for new geoportals has been added.

The map object classifier file access component (axMapRsc) has been improved. The new version adds support for "Date and Time" semantics, enabling input and processing of various date and time formats when working with spatial databases, as well as performing object search and selection operations based on specified conditions. Semantic values are stored in Greenwich Mean Time and displayed in local time with time zone indication on client workstations (e.g., "01/01/2025 12:57:09 +3").

In the "Libraries" tab under the "Formulas" section, the ability to view the list of formulas connected to the iml-library classifier has been implemented. When selecting a formula, the list of classifier semantics with the "Programmable Formula" type is populated. For the selected semantics, corresponding objects are displayed as a list of symbols or a table with their characteristics.

When forming the object table as a list, the ability to configure object property display has been added. The table can be sorted by object property in alphabetical ascending or descending order. By default, the object table is displayed as a matrix of conventional signs.

Support for semi-transparent polygon area figures has been implemented. To select a new type, users must choose "Area" in the elements list, select "Area-glass" in the "View" tab, and enter a transparency value.

The geoportal connection component (axMapWMSPopupDialog) has been enhanced. Support for vector data in GeoPackage (GPKG) format has been added — a modern, open, and cross-platform format developed for storing and exchanging geospatial information. GeoPackage allows storing vector data, metadata, and attribute information in a single SQLite database file.

GIS ToolKit Active contains a set of visual and non-visual components enabling the use of spatial and attribute data for display and special calculations. These components are used to create geoinformation systems at various levels (federal, regional, municipal, corporate). The toolkit implements support for local, state, and international coordinate systems (parameters for over 4,000 different coordinate systems are included in the delivery package). An extensive Geospatial Data Bank is available for use with the toolkit.

The new version of the program is available for download on the Download page.

Geoportal "Bank of spatial data" has been supplemented with a map of the Kingdom of Thailand, formed according to OpenStreetMap data. To work with the map, you can use the GIS Panorama Mini program or other software products of KB "Panorama".

The map was led to the modernised classifier of large-scale plans of scale 1: 5 000 (map5000m). Publishing of updates in bank of spatial data is made by using the program of Databank of digital maps and remote sensing data. Publishing contents of bank of spatial data on the geoportal and data access for downloading are implemented by means of GIS WebServer SE.

Free maps on the basis of OpenStreetMap data are available for download on the page "Digital maps".

KB "Panorama" has developed the automatic recognition and vectorization service "Panorama Vision" version 1.1.1. The new version features improved accuracy in interpreting building contours and agricultural plots. The complex integrates a new neural network model optimized for recognizing building contours in terrain images with spatial resolution of 0.1-0.3 m/pixel. This model enhances building recognition quality on orthophotomaps generated by "Panorama Photo" software from aerial photography processing.

The combined use of "Panorama Photo" and "Panorama Vision" software products maximizes automation in generating ready-to-use geodata. This includes orthophotomaps, elevation matrices, and vector maps created from UAV (Unmanned Aerial Vehicle) data and satellite imagery. Additionally, "Panorama Vision" has improved building contour detection on satellite images with spatial resolution of 0.75 m/pixel and lower. For all image categories, algorithms for separating closely positioned buildings have been enhanced during vector map creation. In agricultural land boundary interpretation, automatic "snapping" of land contours to rectangular shapes has been added. When creating vector maps, agricultural land objects, hydrography, and buildings are formed using the unified classifier map5000m.rscz. For agricultural land objects, the "Arable land" object with code 71324100 is used, for hydrography - "Water body" with code 31120002, and for buildings - "Residential building" with code 44200000.

"Panorama Vision" processes terrain images in RSW/JPG/TIFF/PNG formats and data from open geoportals. The software complex recognizes contours of field plots, hydrographic objects (rivers, ponds, lakes, etc.), buildings, and structures. Processing results are provided to users as vector maps, accessible via WFS protocol. The complex is implemented as services providing remote access to data interpretation. It is hosted on KB "Panorama" servers on the Internet. Access to the complex is available through the GIS "Panorama" application list: Launch applications - Automatic object creation option.

The current version of GIS "Panorama" is available for download in the Download section.

KB "Panorama" has developed GIS "Panorama" version 15.3.3 for "Astra Linux Special Edition" OS. The new version features enhanced tools for contour line generation from point objects, matrix cutting, and elevation matrix alignment. Tasks for transferring and calculating semantics of incoming objects have been updated.

The "Matrix Cutting" mode has been improved, enabling automatic generation of separate matrix map files in MTW format. The task ensures precise data cutting from one or several matrices considering the dimensions of selected objects. Point elevations are determined using the most accurate source matrix overlapping the selected area.

Users can specify names for created files using sequential numbers or object references. Features include file compression for disk space saving, setting matrix boundaries by selected object, and excluding elements outside the specified frame. Additionally, the task allows selecting and adjusting matrix frames while evaluating image dimensions and future file size. This mode automatically controls selection correctness, ensuring convenient and accurate data processing.

The "Elevation Matrix Alignment" task has been enhanced, designed for aligning elevation matrices in the work area. Users can select a control mode that compares matrix elevations at points with identical coordinates (x, y), recording results in a protocol. Error threshold settings are available, determining acceptable deviation between elevations. In editing mode, elevation smoothing is performed in overlap areas using weighted average or arithmetic mean calculations, without using the error threshold. Additionally, the ability to save backup copies of source matrices with *.BAK extension is provided.

The "Contour Line Generation" mode has been updated, ensuring accurate display of equal value lines based on picket point sets. Characteristics can be stored in picket point semantics or in a linked database, with users independently determining the composition of processed points and characteristic locations. Point filtering is available both through the map editor before application launch and through task dialog settings. If data is stored in a database table, preliminary linking of picket points with table records is required.

The contour line generation process occurs within point set boundaries, eliminating extrapolation errors outside the data. Contour line formation is performed in several stages: Delaunay triangulation construction; cross-section creation; contour line tracking along sections and subsequent smoothing. The result is a custom map containing the generated contour lines.

The "Transfer of Polygon Semantics to Incoming Objects" mode has been enhanced. This task automates data transfer between objects considering their territorial dependencies and semantic consistency. For successful transfer, data is taken from reference polygon objects and applied to selected objects on the editable map. If map classifiers differ, users can select one characteristic and reconcile it manually. With identical classifiers, multiple selection of copied characteristics is possible. Semantic reconciliation is performed automatically.

During operation, the following parameters must be specified: editable map name; characteristic to be filled; source map; list of reference objects and semantic characteristic to be copied. If the characteristic has a classifier type, decoding mode can be enabled to replace numerical codes with text values. For example, code 50 in the "Subject of RF" semantics will be converted to "Moscow Region". After completion, these settings are saved in an INI file, simplifying tool reuse.

The "Calculation of Semantic Values for Incoming Objects" task has been improved, supporting two operating modes: calculations by semantics and transfer of semantics from incoming objects to selected polygons. The calculation mode allows analyzing object distribution within selected polygons. It computes indicators such as total, average, minimum, and maximum characteristic values, object inclusion percentage, and the number of objects fully or partially belonging to the polygon. Parameters like object area or length can be considered. Analysis results include user-defined polygon semantic characteristics.

In semantics transfer mode, characteristics from incoming objects are copied to polygons considering map classifier compatibility. Users can manually specify required characteristic codes and select value writing operations.

Task execution includes filters by layers, objects, and localization, with the ability to consider object inclusion coefficients in polygons. This enables precise value distribution, proportional to the area or length of intersecting objects. Additionally, a classifier editing tool is available for configuring characteristic repeatability. The task interface includes convenient tools for clearing and editing data in the semantics table.

GIS "Panorama" is a universal domestic geographic information system. The program is designed for accumulation, storage, automated processing, and display of data, calculation results, and forecasts with geospatial reference. GIS "Panorama" contains tools for displaying and processing vector, raster, and matrix maps. The application provides users with tools for solving specialized tasks of modeling special object movement against 3D models, forecasting emergency consequences at hazardous facilities, monitoring aircraft movement, conducting geodetic and topographic work, processing data from GPS/GLONASS navigation devices, printing displayed spatial data to various output devices, and many others.

The new version of the program is available for download in the Download section.