王志明等：基于深度学习的高效火车号识别 ·1533 [3]Liao J.Research on recognition of railway wagon numbers based 06-24)[2019-11-12].https://arxiv.org/abs/1906.09756v1 on deep convolutional neural networks.J Transp Eng Inf,2016, [15]Liu W,Anguelov D,Erhan D,et al.SSD:Single shot multibox 14(4):64 detector /European Conference on Computer Vision.Amsterdam, (廖健.基于深度卷积神经网络的货车车号识别研究.交通运输 2016:21 工程与信息学报，2016,14(4)：64) [16]Redmon J,Divvala S,Girshick R,et al.You only look once: [4]Li H,Wang P,You M Y,et al.Reading car license plates using Unified,real-time object detection /Proceedings of the IEEE deep neural networks.Image Vision Comput,2018,72:14 Conference on Computer Vision and Pattern Recognition.Las [5]Li H,Wang P,Shen C H.Toward end-to-end car license plate Vegas,2016:779 detection and recognition with deep neural networks.IEEE Trans [17]Rezatofighi H,Tsoi N,Gwak J Y,et al.Generalized intersection Intell Transp Syst,2019,20(3):1126 over union:a metric and a loss for bounding box regression //2019 [6] Montazzolli S,Jung C.Real-time Brazilian license plate detection IEEE/CVF Conference on Computer Vision and Pattern and recognition using deep convolutional neural networks //2017 Recognition.Long Beach,2019:2961 30th SIBGRAPI Conference on Graphics,Patterns and Images [18]Zheng Z H,Wang P,Liu W,et al.Distance-loU Loss:faster and (SIBGRAPI).Niteroi,2017:52 better leaming for bounding box regression [J/OL].arXiv preprint [7]Laroca R,Severo E,Zanlorensi L A,et al.A robust real-time (2019-11-19)[2019-12-151.htps://arxiv.org/abs/1911.08287 automatic license plate recognition based on the YOLO detector / [19]Lin T Y,Goyal P,Girshick R,et al.Focal loss for dense object 2018 International Joint Conference on Neural Networks (LCNN). detection.IEEE Trans Pattern Anal Mach Intell,2020,42(2):318 Rio de Janeiro,2018:1 [20]Shen Z Q,Liu Z,Li J G,et al.DSOD:Leaming deeply supervised [8]Zhang Q.Li J F,Zhuo L.Review of Vehicle Recognition object detectors from scratch /Proceedings of the IEEE Technology.J Beijing Univ Technol,2018,44(3):382 International Conference on Computer Vision.Venice,2017:1919 (张强、李嘉锋、卓力.车辆识别技术综述.北京工业大学学报， [21]Law H,Deng J.CornerNet:detecting objects as paired keypoints 2018,44(3):382) [J/OL].arXiv preprint (2019-03-18)[2019-12-15].https://arxiv.org/ [9] Zhao Z Q,Zheng P,Xu S T,et al.Object detection with deep abs/1808.01244v2 learning:a review.IEEE Trans Neural Networks Learning Syst, [22]Duan K W,Bai S,Xie LX,et al.CenterNet:Keypoint triplets for 2019,30(11:3212 object detection [J/OL].arXiy preprint (2019-04-19)[2019-12-15] [10]Ren S Q,He K M,Girshick R,et al.Faster R-CNN:towards real- https://arxiv.org/abs/1904.08189v3 time object detection with region proposal networks.IEEE Trans [23]Rashwan A.Agarwal R,Kalra A,et al.MatrixNets:a new scale Pattern Anal Mach Intell,2017,39(6):1137 and aspect ratio aware architecture for object detection[J/OL]. [11]Lin TY,Dollar P,Girshick R,et al.Feature pyramid networks for arXiy preprint (2020-01-09)[2020-01-15].https://arxiv.org/abs/ object detection I Proceedings of the IEEE Conference on 2001.03194v1 Computer Vision and Pattern Recognition.Honolulu,2017:2117 [24]Redmon J,Farhadi A.YOLOv3:an incremental improvement [12]He K M,Gkioxari G,Dollar P,et al.Mask R-CNN.[J/OL].arXiv [J/OL].arXiv preprint (2018-04-08)[2019-11-12].https://arxiv. preprint(2018-01-24)[2019-12-1].htps://arxiv.org/abs/1703. org/abs/1804.02767 06870 [25]Liu Z,Li J G,Shen Z Q,et al.Learning efficient convolutional [13]Lu X,Li B Y,Yue Y X,et al.Grid R-CNN plus:faster and better networks through network slimming /2017 IEEE International [J/OL].arXiv preprint (2019-06-13)[2019-12-15].https://arxiv. Conference on Computer Vision.Venice,2017:2755 org/abs/1906.05688v1 [26]Chen K,Wang J Q,Pang J M,et al.MMDetection:Open MMLab [14]Cai Z W,Vasconcelos N.Cascade R-CNN:high quality object detection toolbox and benchmark [J/OL].arXi preprint (2019-06- detection and instance segmentation[J/OL].arXi preprint(2019- 17)[2019-11-121.https://arxiv.org/abs/1906.07155v1Liao J. Research on recognition of railway wagon numbers based on  deep  convolutional  neural  networks. J Transp Eng Inf,  2016, 14（4）: 64 （廖健. 基于深度卷积神经网络的货车车号识别研究. 交通运输 工程与信息学报, 2016, 14（4）：64） [3] Li H, Wang P, You M Y, et al. Reading car license plates using deep neural networks. Image Vision Comput, 2018, 72: 14 [4] Li  H,  Wang  P,  Shen  C  H.  Toward  end-to-end  car  license  plate detection and recognition with deep neural networks. IEEE Trans Intell Transp Syst, 2019, 20（3）: 1126 [5] Montazzolli S, Jung C. Real-time Brazilian license plate detection and recognition using deep convolutional neural networks // 2017 30th SIBGRAPI Conference on Graphics, Patterns and Images (SIBGRAPI). Niterói, 2017: 52 [6] Laroca  R,  Severo  E,  Zanlorensi  L  A,  et  al.  A  robust  real-time automatic license plate recognition based on the YOLO detector // 2018 International Joint Conference on Neural Networks (IJCNN). Rio de Janeiro, 2018: 1 [7] Zhang  Q,  Li  J  F,  Zhuo  L.  Review  of  Vehicle  Recognition Technology. J Beijing Univ Technol, 2018, 44（3）: 382 （张强、李嘉锋、卓力. 车辆识别技术综述. 北京工业大学学报, 2018, 44（3）：382） [8] Zhao  Z  Q,  Zheng  P,  Xu  S  T,  et  al.  Object  detection  with  deep learning:  a  review. IEEE Trans Neural Networks Learning Syst, 2019, 30（11）: 3212 [9] Ren S Q, He K M, Girshick R, et al. Faster R-CNN: towards real￾time object detection with region proposal networks. IEEE Trans Pattern Anal Mach Intell, 2017, 39（6）: 1137 [10] Lin T Y, Dollár P, Girshick R, et al. Feature pyramid networks for object  detection  // Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Honolulu, 2017: 2117 [11] He K M, Gkioxari G, Dollár P, et al. Mask R-CNN. [J/OL]. arXiv preprint (2018-01-24)  [2019-12-15]. https://arxiv.org/abs/1703. 06870 [12] Lu X, Li B Y, Yue Y X, et al. Grid R-CNN plus: faster and better [J/OL]. arXiv preprint (2019-06-13)  [2019-12-15]. https://arxiv. org/abs/1906.05688v1 [13] Cai  Z  W,  Vasconcelos  N.  Cascade  R-CNN:  high  quality  object detection and instance segmentation[J/OL]. arXiv preprint (2019- [14] 06-24) [2019-11-12]. https://arxiv.org/abs/1906.09756v1 Liu  W,  Anguelov  D,  Erhan  D,  et  al.  SSD:  Single  shot  multibox detector // European Conference on Computer Vision. Amsterdam, 2016: 21 [15] Redmon  J,  Divvala  S,  Girshick  R,  et  al.  You  only  look  once: Unified,  real-time  object  detection  // Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition.  Las Vegas, 2016: 779 [16] Rezatofighi H, Tsoi N, Gwak J Y, et al. Generalized intersection over union: a metric and a loss for bounding box regression // 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Long Beach, 2019: 2961 [17] Zheng Z H, Wang P, Liu W, et al. Distance-IoU Loss: faster and better learning for bounding box regression [J/OL]. arXiv preprint (2019-11-19) [2019-12-15]. https://arxiv.org/abs/1911.08287 [18] Lin  T  Y,  Goyal  P,  Girshick  R,  et  al.  Focal  loss  for  dense  object detection. IEEE Trans Pattern Anal Mach Intell, 2020, 42（2）: 318 [19] Shen Z Q, Liu Z, Li J G, et al. DSOD: Learning deeply supervised object  detectors  from  scratch  // Proceedings of the IEEE International Conference on Computer Vision. Venice, 2017: 1919 [20] Law H, Deng J. CornerNet: detecting objects as paired keypoints [J/OL]. arXiv preprint (2019-03-18) [2019-12-15]. https://arxiv.org/ abs/1808.01244v2 [21] Duan K W, Bai S, Xie L X, et al. CenterNet: Keypoint triplets for object detection [J/OL]. arXiv preprint (2019-04-19) [2019-12-15]. https://arxiv.org/abs/1904.08189v3 [22] Rashwan  A,  Agarwal  R,  Kalra  A,  et  al.  MatrixNets:  a  new  scale and  aspect  ratio  aware  architecture  for  object  detection[J/OL]. arXiv preprint (2020-01-09)  [2020-01-15]. https://arxiv.org/abs/ 2001.03194v1 [23] Redmon  J,  Farhadi  A.  YOLOv3:  an  incremental  improvement [J/OL]. arXiv preprint (2018-04-08)  [2019-11-12]. https://arxiv. org/abs/1804.02767 [24] Liu  Z,  Li  J  G,  Shen  Z  Q,  et  al.  Learning  efficient  convolutional networks  through  network  slimming  //  2017 IEEE International Conference on Computer Vision. Venice, 2017: 2755 [25] Chen K, Wang J Q, Pang J M, et al. MMDetection: Open MMLab detection toolbox and benchmark [J/OL]. arXiv preprint (2019-06- 17) [2019-11-12]. https://arxiv.org/abs/1906.07155v1 [26] 王志明等： 基于深度学习的高效火车号识别 · 1533 ·