簡易檢索 / 詳目顯示

回結果列表
研究生： Abbey Dale Abellanosa
Abbey Dale Abellanosa
論文名稱： Automated Crack Segmentation for Rapid Inspection on Concrete Pavement Surface
Automated Crack Segmentation for Rapid Inspection on Concrete Pavement Surface
指導教授： 鄭明淵
Min-Yuan Cheng
口試委員: 曾仁杰
Ren-Jye Dzeng
呂守陞
Sou-Sen Leu
高明秀
Minh-Tu Cao
學位類別： 碩士
Master
系所名稱： 工程學院 - 營建工程系
Department of Civil and Construction Engineering
論文出版年： 2019
畢業學年度： 107
語文別： 英文
論文頁數： 78
中文關鍵詞： Pavement Surface Cracks 、Crack Segmentation 、Rapid Assessment 、Deep Learning 、Mask R-CNN
外文關鍵詞： Pavement Surface Cracks, Crack Segmentation, Rapid Assessment, Deep Learning, Mask R-CNN
相關次數： 點閱：326  下載：1 
分享至:
查詢本校圖書館目錄 查詢臺灣博碩士論文知識加值系統 勘誤回報
  •  上一筆

    • Automatic detection and segmentation of civil structural defects remains a challenge. Its full automation is significant for a cost efficient and effective implementation for maintenance management programs. This study tackles to solve the challenging image segmentation task of concrete pavement surface cracks using the robust deep learning algorithm, mask region-based convolutional neural network (Mask R-CNN). Mask R-CNN has the advantage for this application due to its ability to do instance segmentation at localized regions on an image at 5 fps. This algorithm is tested on an existing pavement crack image database shared in the literature. Moreover, the model performance is evaluated with the same metrics as the COCO competition. Positive detections are retrieved at an intersection-over-union (IoU) threshold of 0.50 and below. In addition to that, the model is highly sensitive to the input image resolution. This study investigated the resolution that yields the optimum accuracy. The successful implementation of this study would ease the rapid detection and segmentation of civil structural defects on a single algorithm.

    Automatic detection and segmentation of civil structural defects remains a challenge. Its full automation is significant for a cost efficient and effective implementation for maintenance management programs. This study tackles to solve the challenging image segmentation task of concrete pavement surface cracks using the robust deep learning algorithm, mask region-based convolutional neural network (Mask R-CNN). Mask R-CNN has the advantage for this application due to its ability to do instance segmentation at localized regions on an image at 5 fps. This algorithm is tested on an existing pavement crack image database shared in the literature. Moreover, the model performance is evaluated with the same metrics as the COCO competition. Positive detections are retrieved at an intersection-over-union (IoU) threshold of 0.50 and below. In addition to that, the model is highly sensitive to the input image resolution. This study investigated the resolution that yields the optimum accuracy. The successful implementation of this study would ease the rapid detection and segmentation of civil structural defects on a single algorithm.

    Chapter 1: Introduction 2 1.1 Background 2 1.2 Research Goal and Objectives 5 1.3 Research Scope 5 1.4 Research Methodology 6 Chapter 2: Literature Review 7 2.1 Rapid Assessment on Concrete Pavement Surface 7 2.2 Previous Automated Pavement Crack Detection Methods 10 2.3 Elements of Mask R-CNN 11 2.3.1 Baseline Convolutional Neural Network Model 12 2.3.2 R-CNN 16 2.3.3 Fast R-CNN 18 2.3.4 Faster R-CNN 19 2.3.5 Mask R-CNN 21 Chapter 3: Model Implementation 24 3.1 Implementation Framework 24 3.2 Image Acquisition 25 3.3 Image Preprocessing 26 3.3.1 Image Annotation 26 3.3.2 Image Augmentation 29 3.4 Cracks Image Database 31 3.5 Parameter Configuration 31 3.5.1 Input Image Resolution Selection 31 3.5.2 Mask R-CNN Hyperparameters 32 3.6 Mask R-CNN Model Training 33 3.6.1 Backbone CNN Architecture 33 3.6.2 Training the Network Parameters 35 3.6.3 Monitoring the Loss Functions 35 3.7 Selection of the Best Mask R-CNN Model 35 3.8 Mask R-CNN Inference Model Testing 36 3.9 Performance Evaluation Criteria 36 3.9.1 Intersection Over Union (IoU) 37 3.9.2 Precision and Recall 37 3.9.3 Average Precision (AP) 38 Chapter 4: Pavement Crack Segmentation Validation and Results 39 4.1 Implementation Details 39 4.1.1 Computer Specifications 39 4.1.2 Software 39 4.2 Experimental Data 39 4.2.1 Image Acquisition 39 4.2.2 Image Description 39 4.3 Training Losses and Convergence Analysis 40 4.3.1 Mask R-CNN Mask Loss 41 4.3.2 Mask R-CNN Multi-Task Loss 41 4.4 Test Results 44 4.5 Crack Prediction Visualization 47 Chapter 5: Conclusion 54 References Appendix A: Mask R-CNN Training Losses (128 × 128 pixels) Appendix B: Mask R-CNN Training Losses (254 × 254 pixels) Appendix C: Mask R-CNN Training Losses (512 × 512 pixels)

    1. Al-Falahi, M. and A. Kassim (2019). Automated Data Collection System of Pavement Distresses: Development, Evaluation & Validation of Distress Types and Severities.
    2. Albawi, S., T. A. Mohammed and S. Al-Zawi (2017). Understanding of a convolutional neural network. 2017 International Conference on Engineering and Technology (ICET).
    3. ASTM-D6433 (2009). American Society for Testing and Materials. Standard Practice for Roads and Parking Lots Pavement Condition Index Surveys. United States.
    4. Atorod Azizinamini, Edward Power, Glenn Myers, Celik Ozyildirim, Eric Kline, David W. Whitmore and D. R. Mertz (2013). Design Guide for Bridges for Service Life. F. H. Administration, National Academy of Sciences.
    5. Bandara, N. and M. Gunaratne (2001). "Current and Future Pavement Maintenance Prioritization Based on Rapid Visual Condition Evaluation." Journal of Transportation Engineering 127(2): 116-123.
    6. Bossi, G., L. Schenato and G. Marcato (2017). "Structural Health Monitoring of a Road Tunnel Intersecting a Large and Active Landslide." Appl Sci 2017
    7. Brownjohn, J. M. W. (2007). "Structural health monitoring of civil infrastructure." Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences 365(1851): 589-622.
    8. DA, D. o. t. A. (1982). TM 5-623 Pavement Maintenance Management. Technical Manual. Washington, DC.
    9. Dawood, T., Z. Zhu and T. Zayed (2018). "Computer Vision-Based Model for Moisture Marks Detection and Recognition in Subway Networks." Journal of Computing in Civil Engineering 32(2): 04017079.
    10. Doycheva, K., C. Koch and M. König (2017). "GPU-Enabled Pavement Distress Image Classification in Real Time." Journal of Computing in Civil Engineering 31(3): 04016061.
    11. Fan, Z., Y. Wu, J. Lu and W. Li (2018). "Automatic Pavement Crack Detection Based on Structured Prediction with the Convolutional Neural Network." Computer Vision and Pattern Recognition. arXiv:1802.02208 [cs.CV].
    12. Feng, C., M.-Y. Liu, C.-C. Kao and T.-Y. Lee (2017). Deep Active Learning for Civil Infrastructure Defect Detection and Classification. Computing in Civil Engineering 2017: 298-306.
    13. Ferguson, M., R. Ak, Y. T. Lee and K. H. Law (2017). Automatic localization of casting defects with convolutional neural networks. 2017 IEEE International Conference on Big Data (Big Data).
    14. Girshick, R. (2015). Fast R-CNN. 2015 IEEE International Conference on Computer Vision (ICCV).
    15. Girshick, R., J. Donahue, T. Darrell and J. Malik (2014). Rich Feature Hierarchies for Accurate Object Detection and Semantic Segmentation. 2014 IEEE Conference on Computer Vision and Pattern Recognition.
    16. He, K., G. Gkioxari, P. Dollár and R. Girshick (2017). Mask R-CNN. 2017 IEEE International Conference on Computer Vision (ICCV).
    17. He, K., X. Zhang, S. Ren and J. Sun (2016). Deep Residual Learning for Image Recognition. 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).
    18. Hinton, G., S. Osindero and Y.-W. Teh (2006). A Fast Learning Algorithm for Deep Belief Nets.
    19. Jaderberg, M., K. Simonyan, A. Zisserman and K. Kavukcuoglu (2015). Spatial Transformer Networks. NIPS.
    20. Jung, A. (2018). imgaug, {https://github.com/aleju/imgaug}.
    21. Karim, F. M. A., K. A. H. Rubasi and A. A. Saleh (2016). "The Road Pavement Condition Index (PCI) Evaluation and Maintenance: A Case Study of Yemen." Organization, Technology and Management in Construction 2016.
    22. Krizhevsky, A., I. Sutskever and G. E. Hinton (2012). ImageNet Classification with Deep Convolutional Neural Networks. NIPS'12 Proceedings of the 25th International Conference on Neural Information Processing Systems, Lake Tahoe, Nevada.
    23. Lecun, Y. (2001). A Theoretical Framework for Back-Propagation.
    24. Maeda, H., Y. Sekimoto, T. Seto, T. Kashiyama and H. Omata (2018). Road Damage Detection Using Deep Neural Networks with Images Captured Through a Smartphone.
    25. Mohan, A. and S. Poobal (2018). "Crack detection using image processing: A critical review and analysis." Alexandria Engineering Journal 57(2): 787-798.
    26. Nakahara, H., H. Yonekawa and S. Sato (2017). An object detector based on multiscale sliding window search using a fully pipelined binarized CNN on an FPGA. 2017 International Conference on Field Programmable Technology (ICFPT).
    27. Oliveira, H. and P. L. Correia (2013). "Automatic Road Crack Detection and Characterization." IEEE Transactions on Intelligent Transportation Systems 14(1): 155-168.
    28. Ren, S., K. He, R. Girshick and J. Sun (2017). "Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks." IEEE Transactions on Pattern Analysis and Machine Intelligence 39(6): 1137-1149.
    29. Salman, M., S. Mathavan, K. Kamal and M. Rahman (2013). Pavement crack detection using the Gabor filter. 16th International IEEE Conference on Intelligent Transportation Systems (ITSC 2013).
    30. Tsai, Y.-C., V. Kaul and R. M. Mersereau (2010). "Critical Assessment of Pavement Distress Segmentation Methods." Journal of Transportation Engineering 136(1): 11-19.
    31. Vasconcelos, C. and B. Nader Vasconcelos (2017). Increasing Deep Learning Melanoma Classification by Classical And Expert Knowledge Based Image Transforms.
    32. Vines-Cavanaugh, D., S. S. Shamsabadi, Y. Zhao, G. Huang, S. Wadia-Fascetti, G. Schirner, R. Birken and M. Wang (2017). "City-Wide Application of the Affordable and Rapid StreetScan Pavement-Management System." Journal of Infrastructure Systems 23(2): B4016010.
    33. Yang, F., L. Zhang, S. Yu, D. Prokhorov, X. Mei and H. Ling (2019). "Feature Pyramid and Hierarchical Boosting Network for Pavement Crack Detection." IEEE Transactions on Intelligent Transportation Systems (Under Review).
    34. Zhang, C., C. C. Chang and M. Jamshidi (2019). "Bridge Damage Detection using a Single-Stage Detector and Field Inspection Images." arXiv preprint arXiv:1812.10590.
    35. Zhang, K., H. D. Cheng and B. Zhang (2018). "Unified Approach to Pavement Crack and Sealed Crack Detection Using Preclassification Based on Transfer Learning." Journal of Computing in Civil Engineering 32(2): 04018001.
    36. Zhang, L., F. Yang, Y. D. Zhang and Y. J. Zhu (2016). Road crack detection using deep convolutional neural network. 2016 IEEE International Conference on Image Processing (ICIP).
    37. Zhou, G. D., T. H. Yi and B. Chen (2017). "Innovative Design of a Health Monitoring System and Its Implementation in a Complicated Long-Span Arch Bridge." Journal of Aerospace Engineering 30(2): B4016006.
    38. Zimmermann, R. and J. Siems (2019). "Faster Training of Mask R-CNN by Focusing on Instance Boundaries." arXiv preprint arXiv:1809.07069 [cs.CV].

    無法下載圖示 全文公開日期 2022/08/05 (校內網路)
    全文公開日期 2024/08/05 (校外網路)
    全文公開日期 2024/08/05 (國家圖書館：臺灣博碩士論文系統)
    QR CODE