隨著交通監控需求的增長和自駕車的蓬勃發展，電腦視覺技術在近幾年的智能交通監控系統中得到了廣泛的應用。這些系統希望將交通違規取締自動化，例如違規停車，違反危險品運送規範、任意變換車道。但是，這些領域的研究尚不成熟。例如，大多數車道線偵測的作法，僅適用於自動駕駛視角的圖片;且大多數車種分類的類別（例如汽車，摩托車和卡車）都非常簡略。我們的研究目標是希望能以數據增強、後處理，以及階層式分類技術，補齊上述兩點的不足。我們將車道線偵測模型，應用於攝影機視角，並且自動化交通取締、ADAS（例如車道偏離警告）、ROI推薦以及計算消失點位置等功能。使用我們的車種辨識模型，系統可以將便是結果，統計各種車輛得流量，並用於檢討交通政策與都是計畫的不足。

With the ever-increasing demand of traffic monitoring and rise of self-driving technology, computer vision is widely used in intelligent traffic surveillance systems in recent year. These systems dedicated to automatically detect traffic offenses, such as illegal parking, violation of dangerous goods transportation and change lane at will. However, studies in these domain are in an immature stage. For example, most studies of lane instance segmentation are conducted from self-driving perspective and the classes of most works of vehicle classification are very rough, such as car, motorcycle and truck. In this work, we proposed a novel method for solution. Our methods can be applied to both traffic surveillance perspective data and self-driving perspective data. With our methods, systems can automatically detect vanish points, make suggestions on regions of interesting for traffic surveillance, detect traffic violation, ADAS such as lane deviation warning, and make statistics that involves diverse kind of vehicles to review city plans and traffic policies.

[1] J. Long, E. Shelhamer, and T. Darrell, “Fully convolutional networks for semantic segmentation,” in Proceedings of the IEEE conference on computer vision and pattern recognition, pp. 3431–3440, 2015.
[2] V. Badrinarayanan, A. Kendall, and R. Cipolla, “Segnet: A deep convolutional encoderdecoder architecture for image segmentation,” IEEE transactions on pattern analysis and machine intelligence, vol. 39, no. 12, pp. 2481–2495, 2017.
[3] T. Falk, D. Mai, and R. Bensch, “Unet –deep learning for cell counting, detection, and morphometry,” Nature Methods 16, 6770, 2019.
[4] H. Zhao, J. Shi, X. Qi, et al., “Pyramid scene parsing network,” in Proceedings of the IEEE conference on computer vision and pattern recognition, pp. 2881–2890, 2017.
[5] Z. Wang, W. Ren, and Q. Qiu, “Lanenet: Realtime lane detection networks for autonomous driving,” arXiv preprint arXiv:1807.01726, 2018.
[6] D. Neven, B. De Brabandere, S. Georgoulis, et al., “Towards endtoend lane detection: an instance segmentation approach,” in 2018 IEEE intelligent vehicles symposium (IV), pp. 286–291, IEEE, 2018.
[7] J. Redmon and A. Farhadi, “Yolo9000: better, faster, stronger,” in Proceedings of the IEEE conference on computer vision and pattern recognition, pp. 7263–7271, 2017.
[8] C. N. Silla and A. A. Freitas, “A survey of hierarchical classification across different application domains,” Data Mining and Knowledge Discovery, vol. 22, no. 1, pp. 31–72, 2011.
[9] S. Feng, J. Wang, D. Zhu, et al., “Jiqing expressway dataset.” https://github.com/vonsj0210/
Multi-Lane-Detection-Dataset-with-Ground-Truth, May 2019.
[10] N. Ma, X. Zhang, H.T. Zheng, et al., “Shufflenet v2: Practical guidelines for efficient cnn architecture
design,” in Proceedings of the European conference on computer vision (ECCV), pp. 116–131, 2018.
[11] S. Woo, J. Park, J.Y. Lee, and I. S. Kweon, “Cbam: Convolutional block attention module,” 2018.
[12] M. Tan, R. Pang, and Q. V. Le, “Efficientdet: Scalable and efficient object detection,” CoRR, vol. abs/
1911.09070, 2019.
[13] L. Breiman, “Random forests,” Machine Learning, vol. 45, no. 1, pp. 5–32, 2001.
[14] A. Paszke, A. Chaurasia, S. Kim, and E. Culurciello, “Enet: A deep neural network architecture for real time semantic segmentation,” arXiv preprint arXiv:1606.02147, 2016.
[15] J. Hur, S.N. Kang, and S.W. Seo, “Multilane detection in urban driving environments using conditional random fields,” in 2013 IEEE Intelligent Vehicles Symposium (IV), pp. 1297–1302, IEEE, 2013.
[16] D. Gao, W. Yang, H. Zhou, et al., “Deep hierarchical classification for category prediction in ecommerce system,” arXiv preprint arXiv:2005.06692, 2020.
[17] Y. Liu, G. Luo, and F. Dong, “Convolutional network model using hierarchical prediction and its application in clothing image classification,” in 2019 3rd International Conference on Data Science and Business Analytics (ICDSBA), pp. 157–160, 2019.
[18] A. B. Jung, K. Wada, J. Crall, et al., “imgaug.” https://github.com/aleju/imgaug, 2020. Online; accessed 01Feb2020.
[19] A. Vaswani, N. Shazeer, N. Parmar, et al., “Attention is all you need,” arXiv preprint arXiv:1706.03762, 2017.
[20] C. H. Sudre, W. Li, T. Vercauteren, S. Ourselin, et al., “Generalised dice overlap as a deep learning loss function for highly unbalanced segmentations,” CoRR, vol. abs/1707.03237, 2017.
[21] T.Y. Lin, P. Goyal, R. Girshick, et al., “Focal loss for dense object detection,” in Proceedings of the IEEE International Conference on Computer Vision (ICCV), Oct 2017.
[22] S. Louhichi, M. Gzara, and H. B. Abdallah, “A density based algorithm for discovering clusters with varied density,” in 2014 World Congress on Computer Applications and Information Systems (WCCAIS), pp. 1–6, IEEE, 2014.
[23] D. Reynolds, Gaussian Mixture Models, pp. 659–663. Boston, MA: Springer US, 2009.
[24] F. B. M. of Transportation and Communications, “Special vehicle control information.” https:// www.freeway.gov.tw/Publish.aspx?cnid=520&p=133, 2008.
[25] D. P. Kingma and J. Ba, “Adam: A method for stochastic optimization,” arXiv preprint arXiv:1412.6980, 2014. 59