딥러닝 기반 석탑 손상 자동 탐지 및 시각화 시스템 개발
Development of Automatic Detection and Visualization System for Stone Pagoda Damage Based on Deep Learning
1
조회
0
다운로드
0
인용
0
참고문헌
키워드
초록
모니터링 기반 손상 진단은 체계적이고 효율적인 문화유산의 보존 관리를 위해 필수적이다. 이를 위해 문화유산의 상시 모니터링이 가능하도록 석탑의 손상을 자동으로 탐지하고시각화하는 딥러닝 시스템을 제안한다. 석탑 이미지에서 손상을 픽셀 단위로 탐지하고 시각화하기 위해 Mask R-CNN을 활용하였으며, 우리나라 석탑에 특화된 데이터세트를 구축하여 딥러닝 모델을 훈련하였다. 훈련된 모델을 이용하여 정림사지 오층 석탑을 대상으로 성능을 평가하였다. 각 유형별 손상 탐지 재현율은 IoU 0.50 기준 0.62부터 0.86의 범위에 있었으며, 손상 영역분할 재현율은 0.51부터 0.68의 범위로 나타났다. 이러한 결과는 인공지능을 문화유산 현장에적용하여 새로운 안전관리 방법론을 제시하며 문화유산 보존에 있어 중요한 응용 가능성을가진다.
Abstract
Damage diagnosis through monitoring is essential for systematic and efficientconservation and management of cultural heritage. In this study, we developed a deep learning systemthat automatically detects and visualizes damage to stone pagodas to enable regular monitoring ofcultural heritage. Mask R-CNN was used to detect and visualize damage in pixel units in stone pagodaimages. A dataset specialized for stone pagodas in Korea was built and applied to train the model. The generalized performance of the trained model was evaluated on the five-story stone pagoda atJeongnimsa Temple Site. The damage detection recall for each type was in the range of 0.86 to 0.62based on IoU 0.50, and the damage area segmentation recall was in the range of 0.68 to 0.51. Thisstudy suggests a new safety management methodology by applying artificial intelligence to culturalheritage sites and has important applications in cultural heritage preservation.
참고문헌 (15)
- National Treasure No.9 Buyeo Jeongnimsa Five-story Stone Pagoda Conservation Treatment Report, "Buyeo-gun", ~ (2007)
- Research service for the establishment of a law on the safety and maintenance of tentatively named cultural properties, "Cultural Heritage Administration", 23~ (2012)
- Choi, J., "A study on the installation and operation of cultural property monitoring system based on low power IoT sensors", Journal of the Society of Cultural Heritage Disaster Prevention, 3(2), pp.101~105 (2018)
- Fitzner, Bernd, "Damage diagnosis at stone monuments-weathering forms, damage categories and damage indices", Acta-Universitatis Carolinae Geologica, 1, pp.12~13 (2001)
- He, K., "Mask R-CNN", Proceedings of the IEEE International Conference on Computer Vision, pp.2961~2969 (2017)
- Kim, S. D., "Trends in materials for repair and restoration of cultural properties-Especially focusing on synthetic resins", MUNHWAJAE Korean Journal of Cultural Heritage Studies, 32, pp.185~201 (1999)
- 김성한, "대전 계족산성 원형성벽의 디지털기록화 및 단기모니터링 연구", 자원환경지질, 52(2), pp.169~188 (2019), 대한자원환경지질학회[DOI]
- 김영택, "부여 정림사지 오층석탑의 보존과학적 풍화훼손도 평가", 자원환경지질, 38(6), pp.675~687 (2005), 대한자원환경지질학회
- LeCun, Y., "Backpropagation applied to handwritten zip code recognition", Neural Computation, 1(4), pp.541~551 (1989)
- Ren, S., "Faster R-CNN : Towards real-time object detection with region proposal networks", Advances in Neural Information Processing Systems, 28 (2015)
- "Sports and Tourism Korea Policy Briefing"[URL]
- 2019Key Management Target Cultural Heritage Monitoring Result Report, "National Research Institute of Cultural Heritage", ~ (2020)
- Yan Zhu, "Semantic amodal segmentation", Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (2017)
- Yu, B. M., "Damage degree and nondestructive diagnosis of the five-story stone pagoda of Jeongnimsa Temple site", Master, pp.2016
- Zhu, Mu, "Recall, precision and average precision", pp.230~ (2004), Department of Statistics and Actuarial Science, University of Waterloo