基于改进YOLOv5n-LPRNet的低照度车牌识别方法
作者:
作者单位:

1.国网新疆电力有限公司博尔塔拉供电公司;2.华北电力大学

中图分类号:

TP391.4

基金项目:

国网新疆电力有限公司科技项目(5230BJ230003)

  • 摘要
  • | |
  • 访问统计
  • |
  • 参考文献 [31]
  • | |
  • 引证文献
  • | |
  • 文章评论
    摘要:

    在电动汽车充电场所,为防止非电动汽车占用充电车位可以在充电桩上设置摄像头并结合目标识别技术实现对汽车类型与车牌号的识别,然而该目标识别任务在更复杂的条件如低照度环境下保证识别精度将具有一定的挑战性。为了解决上述问题,本文提出一种轻量化的、易部署于边缘计算设备的、基于改进YOLOv5n-LPRNet的低照度车牌识别方法。该方法的主要思想为增强-分割-识别,通过将CLAHE及GAMMA变换、YOLOv5n分割网络和LPRNet字符识别网络结合起来,实现端到端的车牌识别。具体改进为:运用“低FLOPs陷阱”思想,将YOLOv5n骨干网络中的CBS模块替换为DynamicConv,并将骨干网络中的C3模块与DynamicConv结合;将YOLOv5n颈部网络中的C3模块替换为YOLOv9提出的RepNCSPELAN模块;在LPRNet网络的两个Dropout层后加入EMA注意力机制。实验结果表明,改进后的模型与原模型相比,分割模型的mask类平均精度提升了约2%,同时在保持实时性的前提下损失少量帧数;识别模型的准确率提高了约9%。

    Abstract:

    In electric vehicle charging facilities, to prevent non electric vehicles from occupying charging spaces, cameras can be installed on charging piles and combined with target recognition technology to achieve recognition of vehicle type and license plate number. However, ensuring recognition accuracy in more complex conditions such as low lighting environments will be challenging. The manual method is time-consuming and laborious, with its efficiency not guaranteed. In order to solve the above problems, this paper proposes a lightweight low-light license plate recognition method based on improved YOLOv5n-LPRNet, which can be easily deployed in edge computing devices. The main idea of this method is Enhance-Segmentation-Recognition, which achieves end-to-end license plate recognition by combining CLAHE-GAMMA transform, YOLOv5n segmentation network and LPRNet character recognition network. The specific improvement is as follows: Using the idea of "Low FLOPs pitfall", the CBS module in YOLOv5n backbone network is replaced by DynamicConv, and the C3 module in backbone network is combined with DynamicConv; The C3 module in the neck network of YOLOv5n was replaced by the RepNCSPELAN module proposed by YOLOv9. The EMA attention mechanism is added after two Dropout layers of the LPRNet network. The experimental results show that compared with the original model, the mask_mAP of the improved model is improved by about 2%, and a small number of frames are lost while maintaining real-time performance. The accuracy of the recognition model improved by about 9%.

    参考文献
    [1] 段峥祺.基于图像识别算法的电动汽车的自动充电接口的识别与定位及其控制方法[D].厦门大学,2017.
    [2] 欧巧凤,肖佳兵,谢群群,等.基于深度学习的车检图像多目标检测与识别[J].应用科学学报,2021,39(06):939-951.
    [3] 徐乐先,陈西江,班亚,等.基于深度学习的车位智能检测方法[J].中国激光,2019,46(04):230-241.
    [4] 孙健,宋茂星,邱果,等.基于电动汽车大数据的多等级充电站选址与服务能力研究[J].中国公路学报,2024,37(04):48-60.
    [5] 陈亚临,杨涌文,赵一涵.面向需求响应的电动汽车-充电桩负荷聚合调度优化策略[J/OL].上海电力大学学报,1-6[2024-07-04].
    [6] 杨秀璋,武帅,任天舒,等.基于改进图像增强及CNN的复杂环境车牌识别算法[J].计算机科学,2024,51(S1):574-580.
    [7] 台骋.基于深度学习的车牌识别方法的设计[D].大连海洋大学,2023.
    [8] Long J, Shelhamer E, Darrell T. Fully convolutional networks for semantic segmentation[C]//Proceedings of the IEEE conference on computer vision and pattern recognition. 2015: 3431-3440.
    [9] Ronneberger O, Fischer P, Brox T. U-net: Convolutional networks for biomedical image segmentation[C]// Medical image computing and computer-assisted intervention 2015: 18th international conference, Munich, Germany, October 5-9, 2015, proceedings, part III 18. Springer International Publishing, 2015: 234-241.
    [10] Chen L C, Papandreou G, Kokkinos I,et al.Semantic Image Segmentation with Deep Convolutional Nets and Fully Connected CRFs[J].Computer Science, 2014(4):357-361.DOI:10.1080/17476938708814211.
    [11] Chen L C, Papandreou G, Kokkinos I,et al.DeepLab: Semantic Image Segmentation with Deep Convolutional Nets,Atrous Convolution, and Fully Connected CRFs[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2018,40(4):834-848.? DOI:10.1109/TPAMI.2017.2699184.
    [12] Chen L C, Papandreou G, Schroff F,et al.Rethinking Atrous Convolution for Semantic Image Segmentation[EB/OL]. (2017-12-5) [2024-1-8]. https://arxiv.org/pdf/1706.05587.pdf
    [13] Chen L C , Zhu Y , Papandreou G ,et al.Encoder-Decoder with Atrous Separable Convolution for Semantic Image Segmentation[C]//European Conference on Computer Vision.Springer, Cham, 2018.DOI:10.1007/978-3-030-01234-2_49.
    [14] 周勇,刘泓滨,侯亚东.复杂城市交通场景下的自动驾驶语义分割方法[J/OL].电子测量与仪器学报,1-9[2024-0702].http://kns.cnki.net/kcms/detail/11.2488.TN.20240507.1544.009.html.
    [15] 黄应华,董振川,李昊,等.城市竣工测绘典型要素语义分割PointNet++深度学习模型适用性分析[J].测绘通报,2024,(02):85-89.
    [16] 王金祥,付立军,尹鹏滨,等.基于CNN与Transformer的医学图像分割[J].计算机系统应用,2023,32(04):141-148.
    [17] 荆涛,王仲.光学字符识别技术与展望[J].计算机工程,2003,(02):1-2+80.
    [18] 陈炳权,汪政阳,夏蓉,等.基于轻量级AlexNet网络的秦简文字识别算法[J].中南大学学报(自然科学版),2023,54(09):3506-3517.
    [19] 高尚,李艳玲,葛凤培,等.基于改进卷积神经网络的身份证信息识别[J].计算机工程与设计,2023,44(11):3447-3454.
    [20] 顾允迪,徐望明,何钦.字轮式仪表智能图像抄表系统的设计[J].液晶与显示,2023,38(07):985-996.
    [21] Zuiderveld K .Contrast Limited Adaptive Histogram Equalization[J].Graphics Gems, 1994: 474-485. DOI:10.1016/B978-0-12-336156-1.50061-6.
    [22] Han K, Wang Y, Guo J, et al. ParameterNet: Parameters Are All You Need for Large-scale Visual Pretraining of Mobile Networks[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2024: 15751-15761.
    [23] Chen Y, Dai X, Liu M, et al. Dynamic convolution: Attention over convolution kernels[C]//Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2020: 11030-11039.
    [24] Hu J, Shen L, Sun G. Squeeze-and-excitation networks [C]//Proceedings of the IEEE conference on computer vision and pattern recognition, 2018: 7132-7141.
    [25] Wang C Y, Yeh I H, Liao H Y M. Yolov9: Learning what you want to learn using programmable gradient information[J]. arXiv preprint arXiv:2402.13616, 2024.
    [26] Wang C Y, Bochkovskiy A, Liao H Y M. YOLOv7: Trainable bag-of-freebies sets new state-of-the-art for real-time object detectors[C]//Proceedings of the IEEE/ CVF conference on computer vision and pattern recognition,2023: 7464-7475.
    [27] Zherzdev S, Gruzdev A. LPRNet: License Plate Recognition via Deep Neural Networks[J/OL].(2018-06-27)[2020-6-14]. https://doi.org/10.48550/arxiv.1806.10447.
    [28] Ouyang D, He S, Zhang G, et al. Efficient multi-scale attention module with cross-spatial learning[C]//ICASSP 2023-2023 IEEE International Conference on Acoustics, Speech and Signal Processing. IEEE, 2023: 1-5.
    [29] Zhao H, Shi J, Qi X, et al. Pyramid scene parsing network[C]//Proceedings of the IEEE conference on computer vision and pattern recognition,2017: 2881-2890.
    [30] Xie E, Wang W, Yu Z,et al. SegFormer: Simple and Efficient Design for Semantic Segmentation with Transformers[J]//Advances in Neural Information Processing Systems, 2021,34:12077-12090.
    [31] Ma L, Ma T, Liu R, et al. Toward fast, flexible, and robust low-light image enhancement[C]//Proceedings of the IEEE/CVF conference on computer vision and pattern recognition,2022: 5637-5646.
    相似文献
    引证文献
    引证文献 [0] 您输入的地址无效!
    没有找到您想要的资源,您输入的路径无效!

    网友评论
    网友评论
    分享到微博
    发 布
引用本文
分享
文章指标
  • 点击次数:2
  • 下载次数: 0
  • HTML阅读次数: 0
  • 引用次数: 0
历史
  • 收稿日期:2024-11-11
  • 最后修改日期:2024-12-10
  • 录用日期:2024-12-13
文章二维码
×
《国外电子测量技术》
2025年投稿方式有变