Semantic Segmentation 알고리즘을 이용한 드론 영상 기반 도로 추출 및 변화 탐지

Abstract

Cities are becoming more complex due to rapid industrialization and population growth in modern times. In particular, urban areas are rapidly changing due to housing site development, reconstruction, and demolition. Thus accurate change information is necessary for various purposes, such as High Definition Map for autonomous car driving. In the case of Korea, accurate spatial information can be generated by making a map through the existing map production process. However, when targeting a large area, it has a limitation in that it takes a lot of time and money. Road, one of the map elements, is a hub and essential means of transportation that provides many different resources for human civilization. Road information is used for traffic management, urban planning, road monitoring, GPS navigation, and map updates. Therefore, it is important to update road change information accurately and quickly. To achieve this goal, this study uses Semantic Segmentation algorithms LinkNet, D-LinkNet, and NL-LinkNet to extract roads from drone images and then apply hyperparameter optimization to models with the highest performance. As a result, the LinkNet model using pre-trained ResNet-34 as the encoder achieved 85.125 mIoU. After detecting roads, change detection was performed by comparing the original road lines in 2019 with the results of road extraction inference in the drone image in 2021. Window and threshold were applied for precise change detection, and as a result, F1-score showed the highest performance with 0.75966 when a window size of 16×16 and threshold 25% was applied. Subsequent studies will be conducted to compare the results of this study with those of studies using state-of-the-art object detection algorithms or semi-supervised learning-based Semantic Segmentation techniques. The results of this study are expected to be used to improve the speed of the existing map update process, which was carried out manually for the changed area, through speedy detection of the changed area.