道路標線具有警告、禁制和引導交通等功能，然而在歷經長期使用後會因為褪色或磨損，降低用路人對其之辨識能力，進而影響交通安全導致事故的發生。因此，確保道路標線之清晰與完整至關重要。現行的道路標線養護作業──道路巡查，仰賴人員定期前往實地進行排查，這需要大量的時間及人力投入。為解決此問題，本研究提出利用搭載有全球導航衛星系統/慣性導航系統(Global Navigation Satellite System/Inertial Navigation System, GNSS/INS)、光達(Light Detection And Ranging, LiDAR)和相機感測器之移動測繪系統(Mobile Mapping System, MMS)──智慧巡查車，蒐集並記錄道路資訊，透過一基於多個深度學習模型的流程自動化評估道路標線磨損程度。研究流程可分為四個部分：第一部分，使用一U-Net架構組合EfficientNet-B7編碼器(Encoder)之深度學習模型EffUNet_Marking針對智慧巡查車所拍攝道路影像，萃取道路標線。第二部分，則是使用另一同第一部分架構的深度學習模型EffUNet_Surface針對智慧巡查車所拍攝道路影像，萃取道路路面。第三部分，藉由車載相機外方位參數將高精地圖(High Definition Map, HD Map)標線相關向量圖層透過攝影測量技術投影至像空間，作為道路標線磨損之參考依據。最後，第四部分，以萃取的道路標線、萃取之道路路面與投影的參考標線影像輸入一基於卷積神經網絡(Convolutional Neural Network, CNN)建構之回歸模型EffNet_Worn，並輸出對應特定位置範圍的標線磨損率，能夠用於表示該處20公尺路長標線磨損之程度。此流程在EffUNet_Marking及EffUNet_Surface上IoU (Intersection over Union)可達0.9，而EffNet_Worn所預測的標線磨損率R^2亦可達0.7，說明本研究所提出之流程，不僅能在減輕巡查人員負擔下提升原道路巡查量能，更足以自動化地應付實務上各種複雜道路巡查場景，提供標線磨損率數值協助有關單位快篩標線現況，做成養護標線的決策同時，亦促進交通管理和交通安全。

It is an essential task in ensuring road marking conditions that could affect traffic safety through road inspection. Currently, road inspection referring to road marking relies on vision-based on-site approaches, which is not only time-consuming but also labor-intensive. In this study, a four-section deep learning workflow is proposed utilizing a Mobile Mapping System (MMS) named “Smart Inspection Vehicle” that collects data on the road. The sections in the workflow include segmenting road markings as well as road surfaces from the gathered images using two models EffUNet_Marking and EffUNet_Surface respectively, projecting the road marking vector layer in High Definition Map (HD Map) onto the camera frame as a reference, and employing a regression model based on Convolutional Neural Network (CNN) denoted as EffNet_Worn to determine the worn rate of the road marking from the extracted features. This automated process could help enhance the conventional road inspection while reducing the time and labor efforts, improving the road marking maintenance work by reaching 0.9 IoU (Intersection over Union) at road markings and road surfaces segmentation also 0.7 R^2 regarding the worn rate assessment.

台灣資通產業標準協會(2020)。高精地圖檢核及驗證指引。
交通部(2023)。111年道路交通安全年報。
交通部；內政部(2023)。道路交通標誌標線號誌設置規則。
交通部公路局(2023)。公路養護手冊。
交通部運輸研究所(1996)。交通管制設施規劃與設計手冊──交通標線篇。
李佳昱(2023)。結合高精地圖與移動式測繪系統評估道路標線磨損程度。國立成功大學測量及空間資訊學系碩士論文。
American Society for Testing and Materials. (2020). ASTM D913-20 Standard Practice for Evaluating Degree of Pavement Marking Line Wear.
Babić, D., & Ščukanec, A. (2016). Determining the correlation between daytime and night-time road markings visibility. The Baltic Journal of Road and Bridge Engineering, 11(4), 283–290.
Babić, D., Fiolić, M., Babić, D., & Gates, T. (2020). Road markings and their impact on driver behaviour and road safety: A Systematic review of current findings. Journal of Advanced Transportation, 2020, 1–19.
Biber, P., & Strasser, W. (2003). The normal distributions transform: a new approach to laser scan matching. Proceedings 2003 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS 2003) (Cat. No.03CH37453), Las Vegas, NV, USA.
Bronuela-Ambrocio, L. B., Antes, E. R. D., & De Leon, M. D. S. (2023). Assessment of Longitudinal Pavement Marking Conditions Based on Deep Learning. PHILIPPINE ENGINEERING JOURNAL, 44(1), 65–88.
Chang, C., & Lin, C. (2012). An Efficient Method for Lane-Mark Extraction in Complex Conditions. 2012 9th International Conference on Ubiquitous Intelligence and Computing and 9th International Conference on Autonomic and Trusted Computing, Fukuoka, Japan.
Cruz, M., Klein, A., & Steiner, V. (2016). Sustainability assessment of road marking systems. Transportation Research Procedia, 14, 869–875.
E. Takeuchi and T. Tsubouchi. (2006). A 3-D Scan Matching using Improved 3-D Normal Distributions Transform for Mobile Robotic Mapping, 2006 IEEE/RSJ International Conference on Intelligent Robots and Systems, Beijing, China.
Highways England. (2020). Road Layout Inspection & Assessment: CS 126 - Inspection and assessment of road markings and road studs.
Hu, J., Shen, L., & Sun, G. (2018). Squeeze-and-Excitation Networks. 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition, Salt Lake City, UT, USA, 7132–7141.
Huang, X., Wang, P., Cheng, X., Zhou, D., Geng, Q., & Yang, R. (2020). The ApolloScape Open dataset for autonomous driving and its application. IEEE Transactions on Pattern Analysis and Machine Intelligence, 42(10), 2702–2719.
J. Deng, W. Dong, R. Socher, L. -J. Li, Kai Li and Li Fei-Fei. (2009). ImageNet: A large-scale hierarchical image database. 2009 IEEE Conference on Computer Vision and Pattern Recognition, Miami, FL, USA.
Kong, W., Zhong, T., Mai, X., Zhang, S., Chen, M., & Lv, G. (2022). Automatic Detection and Assessment of Pavement Marking Defects with Street View Imagery at the City Scale. Remote Sensing, 14(16), 4037.
Liu, R., Wang, J., & Zhang, B. (2019). High Definition Map for Automated Driving: Overview and analysis. Journal of Navigation, 73(2), 324–341.
Miller, T. (1992). Benefit-Cost Analysis of Lane Marking. Transportation Research Record, 1334, 38–45.
Nguyen, S. D., Tran, V. P., Tran, T. S., Lee, H. J., & Flores, J. M. (2023). Automated segmentation and deterioration determination of road markings. Journal of Transportation Engineering. Part B. Pavements., 149(3).
Ohio Department of Transportation. (2023). Construction and Material Specifications.
Perlin, K. (1985). An image synthesizer. Computer Graphics, 19(3), 287–296.
Qin, B., Liu, W., Shen, X., Chong, Z. J., Bandyopadhyay, T., Ang, M. H., Frazzoli, E., & Rus, D. (2013). A general framework for road marking detection and analysis. 16th International IEEE Conference on Intelligent Transportation Systems (ITSC 2013), The Hague, Netherlands.
Ronneberger, O., Fischer, P., Brox, T. (2015). U-Net: Convolutional Networks for Biomedical Image Segmentation. Medical Image Computing and Computer-Assisted Intervention – MICCAI 2015, 234–241.
Russakovsky, O., Deng, J., Su, H., Krause, J., Satheesh, S., Ma, S., Huang, Z., Karpathy, A., Khosla, A., Bernstein, M., Berg, A. C., & Fei-Fei, L. (2015). ImageNet Large Scale Visual Recognition Challenge. International Journal of Computer Vision, 115(3), 211–252.
Sandler, M., Howard, A., Zhu, M., Zhmoginov, A., & Chen, L. (2018). MobileNetV2: Inverted Residuals and Linear Bottlenecks. 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition, Salt Lake City, UT, USA, 4510–4520.
Singapore Standard Council. (2022). Specification for hot-applied thermoplastic road marking materials-Materials, performance and application.
Sudre, C.H., Li, W., Vercauteren, T., Ourselin, S., Jorge Cardoso, M. (2017). Generalised Dice Overlap as a Deep Learning Loss Function for Highly Unbalanced Segmentations. Deep Learning in Medical Image Analysis and Multimodal Learning for Clinical Decision Support, 240–248.
Tan, M. and Le, Q.V. (2019) EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks. Proceedings of the 36th International Conference on Machine Learning, 6105–6114.
Wei, C., Li, S., Wu, K., Zhang, Z., & Wang, Y. (2021). Damage inspection for road markings based on images with hierarchical semantic segmentation strategy and dynamic homography estimation. Automation in Construction, 131, 103876.
Xing, Y., Lv, C., Wang, H., Cao, D., & Velenis, E. (2019). Dynamic integration and online evaluation of vision‐based lane detection algorithms. IET Intelligent Transport Systems, 13(1), 55–62.
Xu, S., Wang, J., Wu, P., Shou, W., Wang, X., & Chen, M. (2021). Vision-Based Pavement Marking Detection and Condition Assessment—A Case Study. Applied Sciences, 11(7), 3152.