İnsansız Hava Aracı Görüntülerinden Beton Yüzeyi Çatlaklarının Yolo Mimarileri ile Tespiti

Yıl 2025, Cilt: 12 Sayı: 26, 159 - 172, 31.08.2025

Ebrar Ödübek , Muhammed Enes Atik

https://doi.org/10.54365/adyumbd.1554821

Öz

Köprü, baraj ve kule gibi yapıların sağlamlığının denetlenmesi, sağlıklı ve güvenilir bir çevre ortamının oluşturulmasında kritik öneme sahiptir. Yüzey çatlaklarının tespiti, yapı sağlamlığının denetlenmesinde stratejik öneme sahiptir. Yapı yüzeylerindeki çatlakların hafif ve zararsız görünümü değişen zaman ve hava koşullarıyla daha tehlikeli hale gelebilmektedir. Yüzey çatlaklarının manuel yöntemlerle tespitinin gerçekleştirilmesi insan gücüne dayalı yüksek performans gerektirdiğinden ulaşılması zor ve riskli yapılarda düşük doğruluğa sebebiyet verebilmektedir. Derin öğrenme teknikleriyle yüzey çatlaklarının tespitinin gerçekleştirilmesi, manuel yöntemlerin yaratmış olduğu yüksek maliyet, zaman ve iş gücü problemlerine bir çözüm sunmaktadır. Bu çalışmada tek aşamalı nesne tespitini sağlayan You Only Look Once (YOLO) algoritmasının güncel versiyonlarının beton yüzeylerindeki çatlakları tespit performansları karşılaştırmalı şekilde analiz edilmiştir. Çalışmada, insansız hava aracı (İHA) görüntülerinden oluşan açık kaynak şeklinde sunulan CRACK veri seti kullanılmıştır. YOLOv9s 0.885 mean average precision (mAP) değeriyle en yüksek doğruluğa sahip algoritmadır.

Anahtar Kelimeler

İnsansız hava aracı , Derin öğrenme , Nesne tespiti , YOLO , Beton yüzeyi çatlakları

Detection of Concrete Surface Cracks from Uav Images Using Yolo Architectures

Öz

The inspection of the structural integrity of constructions such as bridges, dams, and towers is critical to creating a healthy and reliable environment. Detecting surface cracks is of strategic importance in monitoring structural integrity. Due to the high performance required for manual detection of surface cracks, it often leads to lower accuracy in hard-to-reach and risky structures, as it relies heavily on human effort. The seemingly mild and harmless appearance of surface cracks on structures can become more dangerous over time and under varying weather conditions. Detecting surface cracks using deep learning techniques has provided a solution to the high cost, time, and labor problems caused by manual methods. In this study, a comparative analysis was conducted on detecting cracks on concrete surfaces using the latest versions of the You Only Look Once (YOLO) algorithm, which enables single-stage object detection. The CRACK dataset, which is publicly available and consists of UAV images, was used in the study. It was concluded that the YOLOv9s algorithm achieved the highest accuracy with a mean average precision (mAP) of 0.885.

Anahtar Kelimeler

Unmanned aerial vehicle , Deep learning , Object detection , YOLO , Concrete surface cracks

Kaynakça

• Woo HJ, Hong WH, Oh J, Baek, SC. Defining Structural Cracks in Exterior Walls of Concrete Buildings Using an Unmanned Aerial Vehicle, Drones 2023; 7(3):149.
• Rahai A, Rahai M, Iraniparast M, Ghatee M. Surface Crack Detection using Deep Convolutional Neural Network in Concrete Structures. In: 2022 IEEE 5th International Conference on Image Processing Applications and Systems (IPAS); 2022; Genova, Italy. doi: 10.1109/IPAS55744.2022.10052790.
• Chen X, Liu C, Chen L, Zhu X, Zhang Y, Wang C. A Pavement Crack Detection and Evaluation Framework for a UAV Inspection System Based Sciences. Appl. Sci. 2024; 14(3): 1157.
• Redmon J, Divvala S, Girshick R, Farhadi A. You only look once: Unified, real-time object detection, In: Proceedings of the IEEE conference on computer vision and pattern recognition; 2016; 779-788.
• Girshick, R. Fast r-cnn. In Proceedings of the IEEE international conference on computer vision; 2015; 1440-1448.
• He K, Gkioxari G, Dollár P, Girshick R. Mask r-cnn. In: Proceedings of the IEEE international conference on computer vision; 2017; 2961-2969.
• Liu W, Anguelov D, Erhan D, Szegedy C, Reed S, Fu CY, Berg AC. Ssd: Single shot multibox detector. In: Computer Vision ECCV 2016: 14th European Conference; 2016; Amsterdam, The Netherlands.
• Qiu Q, Lau D. Real-time detection of cracks in tiled sidewalks using YOLO-based method applied to unmanned aerial vehicle (UAV) images. Autom Constr 2023; 147: 104745.
• Xu X, Zhao M, Shi P, Ren R, He X, Wei X, Yang H. Crack Detection and Comparison Study Based on Faster R-CNN and Mask R CNN. Sensors 2022; 22(3):1215.
• Xing J, Liu Y, Zhang GZ. Improved YOLOV5-Based UAV Pavement Crack Detection. IEEE Sens J 2023; 23(14):15901-15909.
• Tang Z, Chamchong R, Pawara P. A Comparison of Road Damage Detection Based on YOLOv8. In: 2023 International Conference on Machine Learning and Cybernetics (ICMLC); 2023; Adelaide, Australia.
• Silva WRLD, Lucena DSD. Concrete cracks detection based on deep learning image classification. Proceedings 2018; 2(8): 489.
• Yang C, Chen J, Li Z, Huang Y. Structural Crack Detection and Recognition Based on Deep Learning, Appl. Sci. 2021; 11(6): 1-13. https://doi.org/10.3390/app11062868
• Dataset available online: https://universe.roboflow.com/university-bswxt/crack-bphdr/dataset/1 (Erişim Tarihi: 07.05.2025)
• Cepni S, Atik ME, Duran Z. Vehicle detection using different deep learning algorithms from image sequence, Balt. J. Mod. Comput. 2020; 8(2): 347-358.
• Ultralytics. Yolov5, Available online: https://github.com/ultralytics/yolov5 (Erişim tarihi: 29.02.2024)
• Zhang Y, Huang J, Cai F. On Bridge Surface Crack Detection Based on an Improved YOLOv3 Algorithm, IFAC PapersOnLine 2020; 53(2): 8205–8210. https://doi.org/10.1016/j.ifacol.2020.12.1994
• Li X, Li L, Liu Z, Peng Z, Liu S, Zhou S, Chai X, Jiang K. Dam Crack Detection Studies by UAV Based on YOLO Algorithm. In: 2023 2nd International Conference on Robotics, Artificial Intelligence and Intelligent Control (RAIIC); 2023; Mianyang, China. doi: 10.1109/RAIIC59453.2023.10281120.
• Gupta P, Dixit M. Performance Evaluation of Deep Learning Models For Surface Crack Detection. In: 1st International Conference on Innovations in High Speed Communication and Signal Processing (IHCSP); 2023; Bhopal, India. doi: 10.1109/IHCSP56702.2023.10127175.
• Biyik MY, Atik ME, Duran Z. Deep learning-based vehicle detection from orthophoto and spatial accuracy analysis, INT J ENG GEOSCI 2023; 8(2): 138-145.
• Jocher G, Chaurasia A, Qiu J. Ultralytics YOLO (Version 8.0.0) [Computer software]. Available at: https://github.com/ultralytics/ultralytics 2023; (Erişim tarihi: 23.09.2024)
• Nusari ANM, Ozbek IY, Oral EA. Automatic Vehicle Accident Detection and Classification from Images: A Comparison of YOLOv9 and YOLO-NAS Algorithms. In: 32nd Signal Processing and Communications Applications Conference (SIU); 2024; Mersin, Türkiye. doi: 10.1109/SIU61531.2024.10600761.
• Schetakis N, Koutmos V, Papoutsakis N, Stavrakakis K, Stavroulakis GE, Stavrakakis G. Towards geometric digital twins, including damage detection, from photos of residential buildings facades. In: 2024 IEEE International Workshop on Metrology for Living Environment (MetroLivEnv); 2024; Chania, Greece. doi: 10.1109/MetroLivEnv60384.2024.10615756.
• Panduman YYF, Funabiki N, Sukaridhoto S. An Idea of Drone-Based Building Crack Detection System in SEMAR IoT Server Platform. In: IEEE 12th Global Conference on Consumer Electronics; 2023; Nara, Japan. doi: 10.1109/GCCE59613.2023.10315417.
• Hu W, Qin J, Liang S, Jiang Q. A method for concrete crack detection based on improved YOLOv8s. In: Fourth International Conference on Green Communication, Network, and Internet of Things; 2024; Guiyang, China. doi: 10.1117/12.3052752.