Development of the Terrain Change Detection in Construction Sites Using a Deep Learning-Based Instance Segmentation Method

Jong Ho Na; Yoo Seok Jung; Yoon Seuk Oh; Hyu Soung Shin

doi:10.7474/TUS.2025.35.6.726

All Issue

2025 Vol.35, Issue 6 Preview Page Next Page

Original Article

Development of the Terrain Change Detection in Construction Sites Using a Deep Learning-Based Instance Segmentation Method 딥러닝 기반 영상 분석을 통한 건설 현장 지형 변화 검출 기법 개발

31 December 2025. pp. 726-740

PDF XML

Abstract

Footage captured by on-site CCTV was labeled to construct a terrain-object training dataset, and the data were partitioned by time to evaluate the time-series nature of terrain changes. Specifically, the first two months were used for training and validation, and the subsequent two months were reserved for testing. Using a deep learning–based instance segmentation model, the experiments achieved stable detection of terrain objects with an mAP@0.5 of 90.6%, and further confirmed the feasibility of detecting future terrain changes from past image frames. These results demonstrate the potential of image-only approaches to recognize and monitor terrain changes at earthwork sites and lay the groundwork for extensions to long-term time-series forecasting and risk-aware alerting, positioning the method as a core enabling technology for safety monitoring in earthwork operations.

Keywords

Earthwork construction

Terrain object detection

Instance segmentation

Time-series image analysis

본 연구는 토공 공정이 지속적으로 수행되는 대규모 건설현장에서 영상 분석을 통해 지형 변화를 감지하는 기법을 제안한다. 현장 CCTV에서 수집한 영상을 가공하여 지형 객체 학습 데이터셋을 구축하였고, 지형 변화의 시계열성을 평가하기 위해 시점 기준으로 데이터를 분할하였다. 분할 데이터셋의 초기 2개월은 학습 데이터와 검증 데이터로 구성하였고 이후 2개월은 테스트 데이터로 분류하였다. 딥러닝 기반 인스턴스 분할 모델 기반 지형 변화 검출 실험을 수행한 결과, mAP@0.5 90.6% 수준의 정확도로 지형 객체를 안정적으로 검출하였으며, 과거 시점 영상 기반으로 미래 시점의 변화된 지형까지 감지하는 연구 결과를 확인하였다. 본 연구 결과는 영상만으로 토공 현장의 지형 변화를 인식하고 모니터링하는 가능성을 입증하였으며, 향후 장기간 시계열 예측 및 위험도 연계 알림으로 확장하여 토공 현장 안전 모니터링 솔루션의 핵심 요소기술로 활용 될 기반을 마련하였다.

키워드

토공 현장 데이터

지형 객체 검출

인스턴스 분할

시계열 영상 분석

References

Bolya, D., Zhou, C., Xiao, F., and Lee, Y.J., 2019, YOLACT: Real-time instance segmentation, Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), pp. 9157-9166.

10.1109/ICCV.2019.00925

Brostow, G.J., 2019, COCO Annotator.

Cho, H., Lee, S., and Kim, M., 2021, UAV-based earthwork volume estimation using digital surface models, Remote Sensing, 13(3), 512.

Everingham, M., Van Gool, L., Williams, C.K.I., Winn, J., and Zisserman, A., 2010, The PASCAL visual object classes (VOC) challenge, International Journal of Computer Vision, 88(2), 303-338.

10.1007/s11263-009-0275-4

He, K., Gkioxari, G., Dollár, P., and Girshick, R., 2017, Mask R-CNN, in Proceedings of the IEEE International Conference on Computer Vision (ICCV), pp. 2961-2969.

10.1109/ICCV.2017.322

Korea Occupational Safety and Health Agency (KOSHA), 2023, Smart Safety Equipment Testbed Overview.

Lin, T.-Y., Maire, M., Belongie, S., Hays, J., Perona, P., Ramanan, D., Dollár, P., and Zitnick, C.L., 2014, Microsoft COCO: Common objects in context, in Proceedings of the European Conference on Computer Vision (ECCV), pp. 740-755.

10.1007/978-3-319-10602-1_48

Ministry of Employment and Labor (MOEL), 2022, Amendment of the Occupational Safety and Health Act.

Ministry of Employment and Labor (MOEL), 2023, Revised Guidelines for the Use of Occupational Safety and Health Management Expenses.

Ministry of Government Legislation (MGL), 2022, Construction Machinery Management Act.

Ministry of Government Legislation (MOLEG), 2022, Serious Accidents Punishment Act.

Ministry of Land, Infrastructure and Transport (MOLIT), 2014, Development of Risk Factors for Construction Project.

Ministry of Land, Infrastructure and Transport (MOLIT), 2018, Smart Construction Technology Roadmap.

Na, J., Shin, H., Yun, I., and Lee, J., 2025, Development of an AI dataset for object detection at construction sites, Mendeley Data, V2.

Rahman, M.A., and Wang, Y., 2016, Optimizing intersection-over-union in deep neural networks for image segmentation, in Proceedings of the International Symposium on Visual Computing, pp. 234-244.

10.1007/978-3-319-50835-1_22

Rasul, M., Ramalingam, M., and Kumar, R., 2021, 3D point cloud-based excavation monitoring in earthwork construction, Applied Sciences, 11(16), 7356.

Ren, S., He, K., Girshick, R., and Sun, J., 2015, Faster R-CNN: Towards real-time object detection with region proposal networks, in Advances in Neural Information Processing Systems (NeurIPS), 28.

Ren, S., He, K., Girshick, R., and Sun, J. 2016, Faster R-CNN: Towards real-time object detection with region proposal networks. IEEE transactions on pattern analysis and machine intelligence, 39(6), 1137-1149.

10.1109/TPAMI.2016.2577031

Seoul Metropolitan Government, 2023, Announcement of 10 Major Measures to Strengthen Construction Safety.

Wang, X., Kong, T., Shen, C., Jiang, Y., and Li, L., 2020, SOLO: Segmenting Objects by Locations, Computer Vision – ECCV 2020, Lecture Notes in Computer Science, Springer, Cham, 12349, 702-717.

10.1007/978-3-030-58523-5_38

Wu, Y., Kirillov, A., Massa, F., Lo, W.Y., and Girshick, R., 2019, Detectron2, available at: https://github.com/facebookresearch/detectron2

Information

Publisher :Korean Society for Rock Mechanics and Rock Engineering
Publisher(Ko) :한국암반공학회
Journal Title :Tunnel and Underground Space
Journal Title(Ko) :터널과 지하공간
Volume : 35
No :6
Pages :726-740
Received Date : 2025-10-17
Revised Date : 2025-12-19
Accepted Date : 2025-12-22
DOI :https://doi.org/10.7474/TUS.2025.35.6.726

Tunnel and Underground SpaceISSN:1225-1275(Print) 2287-1748(Online)한국암반공학회

All Issue