항공 영상에서의 Mask R-CNN을 이용한 도로 표면 균열 검출 연구

Cracks in roads caused by cyclic loads, drying shrinkage, and changes in temperature can cause damage to the vehicle and traffic accidents if left for a long time. This problem is an essential element in the implementation of GeoAI based urban environment that reflects traffic information. In this paper, we study crack data using Mask R-CNN, which is a deep-running model that is useful for object detection and extraction, and verify crack detection in real aerial images taken by the drone.