目前國內的道路維護方式多為定期派遣工程車檢測以及依賴人民的通報，而為了盡早的發現道路損壞並進行修復，本研究運用Mask R-CNN深度學習之方式建立道路破損辨識模型。透過Mask R-CNN深度學習演算法，以及運用python 、OpenCV撰寫進行道路破損檢測與資料整合，持續的分析模型數據並根據結果進行再訓練。利用路面破損辨識模型檢測出路面上的龜裂、裂縫、補綻、變形以及坑洞，並在龜裂、裂縫、補綻、變形達到86%以上的召回率，精確率除了裂縫、坑洞之外有82%以上，此外對檢測出來的破損範圍進行面積計算，為日後養護維修提供面積的量化指標，進而輔助人力巡查作業。

At present, domestic road maintenance methods are mostly dispatching construction vehicles for inspection and relying on people’s notifications. In order to discover and repair road crack as soon as possible, this study uses Mask R-CNN deep learning to establish a road crack identification model. Through the Mask R-CNN deep learning algorithm, and the use of python and OpenCV to write for road crack detection and data integration, continuous analysis of model data and retraining based on the results. The road surface crack identification model is used to detect alligator crack, linear crack, patch, deformation and pothole on the road, and the recall rate of alligator crack, linear crack, patch and deformation is over 86%. Except for linear crack and pothole, the accuracy rate is more than 82%. In addition, the area of the detected damage area is calculated to provide a quantitative indicator of the area for future maintenance and repair, and then assist manpower inspection operations.

[1] R. Girshick, J. Donahue, T. Darrell, and J. Malik, “Rich feature hierarchies for accurate object detection and semantic segmentation,” in 2014 IEEE Conference on Computer Vision and Pattern Recognition, pp. 580–587, June 2014.
[2] R. Girshick, “Fast r-cnn,” in 2015 IEEE International Conference on Computer Vision (ICCV), pp. 1440–1448, Dec 2015.
[3] S. Ren, K. He, R. Girshick, and J. Sun, “Faster r-cnn: Towards real-time object detection with region proposal networks,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 39, pp. 1137–1149, June 2017.
[4] J. Redmon, S. Divvala, R. Girshick, and A. Farhadi, “You only look once: Unified, real-time object detection,” in 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp. 779–788, June 2016.
[5] K. He, G. Gkioxari, P. Dollar, and R. Girshick, “Mask R-CNN,” CoRR abs/1703.06870 (2017).
[6] Labelme, [Online], Available: https://github.com/wkentaro/labelme
[7] OpenCV, [Online],Available: https://opencv.org
[8] 內政部營建署, “市區道路管理維護與技術規範手冊研究”,2002
[9] K. Fukushima and S. Miyake, “Neocognitron: A self-organizing neural network model for a mechanism of visual pattern recognition,” in Competition and Cooperation in Neural Nets, Springer, pp. 267–285, 1982.
[10] D.H. Hubel and T.N. Wiesel, “Receptive fields and functional architecture of monkey striate cortex,” J. Physiol. (1968) 215–243.
[11] K. Fukushima and S. Miyake, “Neocognitron: a self-organizing neural network model for a mechanism of visual pattern recognition,” in: Competition and Cooperation in Neural Nets, 1982, pp. 267–285.
[12] Y. LeCun, L. Bottou, Y. Bengio and P. Haffner, “Gradient-based learning applied to document recognition,” Proc. IEEE 86 (11) (1998) 2278–2324.
[13] O. Russakovsky, J. Deng, H. Su, J. Krause, S. Satheesh, S. Ma, Z. Huang, A. Karpathy, A. Khosla, M. Bernstein, A. C. Berg, and L. Fei-Fei, “Imagenet large scale visual recognition challenge,” Int. J. Conflict Violence (IJCV) 115 (3) (2015) 211–252.
[14] C. Szegedy, W. Liu, Y. Jia, P. Sermanet, S. Reed, D. Anguelov, D. Erhan, V. Vanhoucke and A. Rabinovich, “Going deeper with convolutions,” in: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2015, pp. 1–9.
[15] K. Simonyan and A. Zisserman, “Very deep convolutional networks for large-scale image recognition,” in: Proceedings of the International Conference on Learning Representations (ICLR), 2015.
[16] K. He, X. Zhang, S. Ren and J. Sun, “Deep residual learning for image recognition,” in: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2016, pp. 770–778.
[17] Image,[Online],Available:https://engineering.matterport.com/7c761e238b46
[18] J. R. Uijlings, K. E. van de Sande, T. Gevers, and A. W. Smeulders, “Selective search for object recognition,” IJCV, 2013. 2.
[19] B. Liu, Z. F. Hao and X. W. Yang, “Nesting Support Vector Machine for Multi-Classification,” ICMLC 2005, pp. 4220-4225, August 2005.
[20] Neubeck A, Gool L V. Efficient Non-Maximum Suppression,”International Conference on Pattern Recognition.” IEEE Computer Society, 2006:850-855
[21] R. Rothe, M. Guillaumin, and L. Van Gool, “Non-maximum suppression for object detection by passing messages between windows" In Proceedings of the Asian Conference on Computer Vision (ACCV), Singapore, 1–5 November 2014.
[22] S. Qiao, L.C. Chen, A. Yuille, “DetectoRS: Detecting Objects with Recursive Feature Pyramid and Switchable Atrous Convolution” IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2021, pp. 10213-10224
[23] L. Tingting, C. Xiaojie, L. Yudong, W. Yongtao, T. Zhi, C. Wei, C. Jingdong and . Haibin, “CBNetV2: A Composite Backbone Network Architecture for Object Detection” arXiv:2107.00420
[24] SuperDataScience, CNN結構示意圖, [Online], Available : SuperDataScience：Convolutional Neural Networks (CNN): Step 3 – Flattening, Aug 2018
[25] jiongnima, Mask R-CNN結構圖, [Online], https://blog.csdn.net/jiongnima/article/details/79094159