Update priority map generation for 3D city model

Abstract

To further promote 3D city models, which contribute to solving problems in diverse areas such as infrastructure maintenance and management and disaster recovery, it is essential to increase the update frequency and maintain the identity with the actual city space. However, increasing the update frequency of 3D city models is difficult because they are maintained using existing survey results based on aerial photographs. Therefore, we developed an AI model that compares the aerial photographs used to create the 3D city model with satellite images, which are frequently taken, to extract changes in buildings, etc., and generated an updated priority map that expresses the difference between the 3D city model and the actual space.