다비오

Dabeeo, a geospatial intelligence technology company, announced that it has been selected by the Korea Asset Management Corporation (KAMCO) to develop an “AI-Based Aerial Imagery Change Detection System for National Property Surveys.” This achievement highlights Dabeeo’s expertise in specialized spatial intelligence and reinforces its position as a trusted partner capable of delivering large-scale public sector projects.

The project aims to enhance the efficiency of nationwide surveys of state-owned property by leveraging AI-powered aerial imagery change detection. By reducing the burden of repetitive annual on-site inspections and concentrating field surveys on areas where changes are detected, the initiative is designed to improve both accuracy and operational efficiency. This effort requires an advanced detection system that overcomes traditional limitations related to seasonal and crop conditions, focusing on accurately identifying real usage changes—a task where AI-driven spatial analytics play a pivotal role.

Using 25 cm-resolution aerial imagery provided by the National Geographic Information Institute, Dabeeo will develop a system capable of analyzing changes in land use across all state-owned properties by comparing imagery captured at different times. The system will detect various changes, including new building construction, unauthorized farming, and illegal storage, while also classifying usage types to support more informed property management.

The project covers the full development process, from training and deploying AI models for change detection to establishing a high-performance infrastructure capable of nationwide analysis within 90 days, and integrating results with KAMCO’s existing platforms such as “GUKYU iN-GIS.” Dabeeo will adopt a Siamese-like model architecture that compares semantic changes in imagery rather than relying solely on classification-based methods, enabling both precision and versatility. The target is to achieve a system with an F1-score accuracy of over 90%.

Additionally, the results will be visualized through a GIS-based intuitive user interface and integrated with a reinforcement learning structure that incorporates feedback on detected changes, accelerating the digital transformation of national property management.

Ju Hum Park, CEO of Dabeeo, stated, “This project represents a pioneering case of applying AI-powered change detection technology to national land management systems and marks an important milestone in advancing digital infrastructure in the public sector. Dabeeo will continue to drive innovation in spatial intelligence technology to enhance policy execution and improve overall public administration efficiency.”