隨著智慧運輸系統的快速發展，自駕車系統成為產官學界共同研究與開發的目標，同時為了建置與維護自駕車用高精地圖，以影像偵測為基礎的交通標誌辨識逐漸成為熱門的研究議題。目前已有許多文獻在公開數據庫訓練交通標誌辨識模型，也在實際道路影像的驗證下，得到高準確度的辨識能力，證實基於影像的交通標誌辨識技術是可行的方案，但這些方法都必須建立在數據庫充足的條件下，目前在臺灣沒有足夠龐大具交通標誌標記的道路影像數據庫，只能利用人工標記的方式產生有限的訓練資料，即使辨識技術已十分成熟也無法被妥善使用。有鑒於此，本論文提出一種半自動化交通標誌的影像標註方法，透過對交通標誌的觀察，使用少量的標註影像來訓練Mask R-CNN，使其依照形狀來偵測交通標誌，再經由影像優化及影像比對來更新類別，此方法只在一開始需要少量的標註影像，之後便能自動化的進行，產生具有交通標誌真實類別的標註數據，協助臺灣建立交通標誌數據庫。透過真實資料的實驗評估，本研究所提出的方法能夠有效提升交通號誌標記的效率，進而大幅改善交通號誌辨識的準確度。

With the rapid development of Intelligent Transportation Systems, autonomous vehicle systems have become the goal of joint research and development by industry, government, and academia. At the same time, to build and maintain High-Definition Map for autonomous vehicles, traffic sign recognition based on image detection has gradually become a popular research topic. In recent years, many studies have been proposed to train a model to detect traffic signs based on public road image databases and obtain a high accuracy on recognition capability under the verification of actual road image, which proves that it is a practical solution to detect traffic signs using the image-based model. However, these methods must be established under enough images in databases. There is currently no large enough road image database with traffic signs in Taiwan, and only limited training data can be generated by manual annotation. Even the detection method is already prepared, it cannot be applied directly. In this thesis, we propose a semi-automatic image annotation method for traffic signs. Through the observation of traffic signs, a small number of annotated images are used to train Mask R-CNN to first detect the shape of a traffic sign. Then, the real label of a traffic sign is predicted by image enhancement and matching. The proposed method only requires a small number of annotated images initially, and the annotated data with traffic sign labels can be automatically generated to establish Taiwan traffic sign databases. Through the experimental evaluation of a real dataset, the proposed method can effectively improve the efficiency of traffic sign annotation and greatly increase the accuracy of traffic sign detection.

[1] Artamonov, N. S., & Yakimov, P. Y. (2018, September). Towards Real-Time Traffic Sign Recognition via YOLO on a Mobile GPU. In Journal of Physics: Conference Series (Vol. 1096, No. 1).
[2] Bahlmann, C., Zhu, Y., Ramesh, V., Pellkofer, M., & Koehler, T. (2005, June). A system for traffic sign detection, tracking, and recognition using color, shape, and motion information. In IEEE Proceedings. Intelligent Vehicles Symposium, 2005. (pp. 255-260). IEEE.
[3] Ćorović, A., Ilić, V., Durić, S., Marijan, M., & Pavković, B. (2018, November). The real-time detection of traffic participants using yolo algorithm. In 2018 26th Telecommunications Forum (TELFOR) (pp. 1-4). IEEE.
[4] Cortes, C., & Vapnik, V. (1995). Support-vector networks. Machine learning, 20(3), 273-297.
[5] Freund, Y., & Schapire, R. E. (1995, March). A desicion-theoretic generalization of on-line learning and an application to boosting. In European conference on computational learning theory (pp. 23-37). Springer, Berlin, Heidelberg.
[6] Girshick, R. (2015). Fast r-cnn. In Proceedings of the IEEE international conference on computer vision (pp. 1440-1448).
[7] Girshick, R., Donahue, J., Darrell, T., & Malik, J. (2014). Rich feature hierarchies for accurate object detection and semantic segmentation. In Proceedings of the IEEE conference on computer vision and pattern recognition (pp. 580-587).
[8] Gómez-Moreno, H., Maldonado-Bascón, S., Gil-Jiménez, P., & Lafuente-Arroyo, S. (2010). Goal evaluation of segmentation algorithms for traffic sign recognition. IEEE Transactions on Intelligent Transportation Systems, 11(4), 917-930.
[9] Gonzalez, R. C., & Woods, R. E. (2002). Digital Image Processing. 2nd edn Prentice Hall. New Jersey, 793.
[10] He, K., Gkioxari, G., Dollár, P., & Girshick, R. (2017). Mask r-cnn. In Proceedings of the IEEE international conference on computer vision (pp. 2961-2969).
[11] Houben, S., Stallkamp, J., Salmen, J., Schlipsing, M., & Igel, C. (2013, August). Detection of traffic signs in real-world images: The German Traffic Sign Detection Benchmark. In The 2013 international joint conference on neural networks (IJCNN) (pp. 1-8). IEEE.
[12] Kiran, C. G., Prabhu, L. V., & Rajeev, K. (2009, February). Traffic sign detection and pattern recognition using support vector machine. In 2009 Seventh International Conference on Advances in Pattern Recognition (pp. 87-90). IEEE.
[13] LeCun, Y., Bottou, L., Bengio, Y., & Haffner, P. (1998). Gradient-based learning applied to document recognition. Proceedings of the IEEE, 86(11), 2278-2324.
[14] Lin, T. Y., Dollár, P., Girshick, R., He, K., Hariharan, B., & Belongie, S. (2017). Feature pyramid networks for object detection. In Proceedings of the IEEE conference on computer vision and pattern recognition (pp. 2117-2125).
[15] Lin, T. Y., Goyal, P., Girshick, R., He, K., & Dollár, P. (2017). Focal loss for dense object detection. In Proceedings of the IEEE international conference on computer vision (pp. 2980-2988).
[16] Liu, Z., Du, J., Tian, F., & Wen, J. (2019). MR-CNN: A multi-scale region-based convolutional neural network for small traffic sign recognition. IEEE Access, 7, 57120-57128.
[17] Lowe, D. G. (1999, September). Object recognition from local scale-invariant features. In Proceedings of the seventh IEEE international conference on computer vision (Vol. 2, pp. 1150-1157). Ieee.
[18] Mathias, M., Timofte, R., Benenson, R., & Van Gool, L. (2013, August). Traffic sign recognition—How far are we from the solution?. In The 2013 international joint conference on Neural networks (IJCNN) (pp. 1-8). IEEE.
[19] Ou, Z., Xiao, F., Xiong, B., Shi, S., & Song, M. (2019). FAMN: Feature aggregation multipath network for small traffic sign detection. IEEE Access, 7, 178798-178810.
[20] Rajendran, S. P., Shine, L., Pradeep, R., & Vijayaraghavan, S. (2019, July). Fast and Accurate Traffic Sign Recognition for Self Driving Cars using RetinaNet based Detector. In 2019 International Conference on Communication and Electronics Systems (ICCES) (pp. 784-790). IEEE.
[21] Redmon, J., Divvala, S., Girshick, R., & Farhadi, A. (2016). You only look once: Unified, real-time object detection. In Proceedings of the IEEE conference on computer vision and pattern recognition (pp. 779-788).
[22] Redmon, J., & Farhadi, A. (2017). YOLO9000: better, faster, stronger. In Proceedings of the IEEE conference on computer vision and pattern recognition (pp. 7263-7271).
[23] Redmon, J., & Farhadi, A. (2018). Yolov3: An incremental improvement. arXiv preprint arXiv:1804.02767.
[24] Ren, S., He, K., Girshick, R., & Sun, J. (2015). Faster r-cnn: Towards real-time object detection with region proposal networks. In Advances in neural information processing systems (pp. 91-99).
[25] Stallkamp, J., Schlipsing, M., Salmen, J., & Igel, C. (2012). Man vs. computer: Benchmarking machine learning algorithms for traffic sign recognition. Neural networks, 32, 323-332.
 
[26] Tabernik, D., & Skočaj, D. (2019). Deep learning for large-scale traffic-sign detection and recognition. IEEE Transactions on Intelligent Transportation Systems, 21(4), 1427-1440.
[27] Takaki, M., & Fujiyoshi, H. (2009). Traffic sign recognition using SIFT features. ITEIS, 129(5), 824-831.
[28] Tran, A. C., Dien, D. L., Huynh, H. X., Van Long, N. H., & Tran, N. C. (2019, November). A Model for Real-Time Traffic Signs Recognition Based on the YOLO Algorithm–A Case Study Using Vietnamese Traffic Signs. In International Conference on Future Data and Security Engineering (pp. 104-116). Springer, Cham.
[29] Uijlings, J. R., Van De Sande, K. E., Gevers, T., & Smeulders, A. W. (2013). Selective search for object recognition. International journal of computer vision, 104(2), 154-171.
[30] Xie, Y., Liu, L. F., Li, C. H., & Qu, Y. Y. (2009, June). Unifying visual saliency with HOG feature learning for traffic sign detection. In 2009 IEEE Intelligent Vehicles Symposium (pp. 24-29). IEEE.
[31] Zhang, J., Huang, M., Jin, X., & Li, X. (2017). A real-time chinese traffic sign detection algorithm based on modified YOLOv2. Algorithms, 10(4), 127.
[32] Zhu, Z., Liang, D., Zhang, S., Huang, X., Li, B., & Hu, S. (2016). Traffic-sign detection and classification in the wild. In Proceedings of the IEEE conference on computer vision and pattern recognition (pp. 2110-2118).