Intelligent Development Potential Analysis for Saxony-Anhalt

Project Overview

As part of the “Intelligent Development Potential Analysis” project, Fraunhofer IGD Rostock developed an AI-based system for identifying vacant lots, infill development sites and areas with industrial development potential from aerial imagery. The aim is to establish a state-wide register of development potential sites for Saxony-Anhalt to support infill development and municipal land-use planning. The automated preselection provides a basis for subsequent assessment by municipal planning experts.

Methodology and Data Basis

The AI analyzes publicly available DOP20 orthophotos (20 cm/pixel), keeping visual detection separate from planning-related assessment. The annotated training dataset comprises
5,150 image sections measuring 350 m × 350 m for residential development potential and
1,000 sections measuring 750 m × 750 m for industrial development potential. In consultation with experts from the municipalities of Magdeburg, Staßfurt and Haldensleben, clear definitions were established for the three types of development potential, along with a standardized annotation strategy that ensures a clear separation between image-based information and planning expertise. Following automated site detection, the results undergo multi-stage post-processing: they are cross-referenced with ALKIS, ATKIS and ROK data, assigned to land parcels and filtered according to land-use categories. Vacant lots and infill development sites are then classified. Industrial development potential is processed by a separate network specifically designed for this type of site. The process is transparent, reproducible and can be rerun whenever new data becomes available.

Results and Evaluation

An independent blind test comprised 400 image sections of built-up areas, each measuring 350 m × 350 m, containing 779 development potential sites validated by the participating municipalities. For residential development potential – vacant lots and infill development sites of 500 m² or more – the system achieves an overall accuracy of 93.6%, with results ranging from 92% to 95% depending on the municipality. For industrial development potential, accuracy is 89.1% per municipality.

Sector and Areas of Application

The project addresses key challenges in municipal urban development and land-use planning. The identified development potential sites provide municipal authorities with suggestions and data-based decision-making support for infill development. This helps limit urban sprawl, preserve agricultural land and natural habitats, and make more efficient use of existing infrastructure.

Added Value and Outlook

The AI-based analysis supports municipalities in the time-consuming initial assessment of large volumes of aerial imagery, significantly improving efficiency and saving time. Existing registers can be reviewed and updated, while previously unidentified sites can be added. Municipalities without their own development land register receive a sound basis for further assessment.

The AI largely automates the initial manual search, while final planning decisions remain the responsibility of the municipalities. The developed pipeline has already been integrated into the Saxony-Anhalt Viewer. Potential future developments include additional exclusion classes, a more extensive dataset for industrial development potential, the integration of XPlanung and other specialist datasets, as well as the evaluation and prioritization of identified sites.