电动车驾驶员的违章行为极易导致电动车与行驶的机动车车辆发生碰撞，威胁未佩戴或未正确佩戴头盔的骑行者与随行人员的生命安全。因此，研究电动车骑行者头盔检测与车牌识别方法具有重大的理论价值和现实意义。但目前人工智能和计算机视觉在智能交通领域的研究对象主要是汽车，缺乏对电动车违章行为和电动车车牌整体的检测与识别。因此，本文针对电动车上述情况展开研究，具体的工作如下：

（1）针对电动车小目标检测准确率低、鲁棒性差、相关系统不完善等问题，本文提出了改进YOLOv5s算法。该改进算法使用了自建的电动车头盔和车牌数据集，通过引入注意力模块CBAM和CA，用NMS-DIOoU改进非极大值抑制NMS，增加多尺度特征融合，实现自动检测电动车头盔佩戴以及车牌的效果。结果表明：改进后的 YOLOv5s比YOLOv5s整体的精确率提高了4.3%，召回率提高了2.7%，平均精确率（IOoU=0.5）提高了3.6%。在检测电动车头盔上更能满足准确性的需求。

（2）针对电动车车牌检测定位精度低，双行车牌识别困难等问题。，本文采用改进 YOLOv5s对电动车数据集进行模型训练来提升车牌检测精度。并通过使用数字图像处理技术对检测到的电动车车牌进行Hough变换矫正，同时结合构建的CRNN的光学字符车牌文字识别模型对电动车车牌文字进行识别，其识别准确率可达到91.3%。在电动车和车牌整体自动检测与识别过程中，单帧处理时间约为109 ms。

（3）针对实际需求，设计电动车违章自动检测系统。该系统实现了包括骑电动车、正确配带头盔、未正确配带头盔、电动车车牌的检测识别、统计计数、帧间延时、图片检测、视频检测等各个模块。该检测系统即可以实现对复杂交通流量下未佩戴安全头盔的电动车违章人员进行检测和统计，帮助执法人员对道路电动车流量大和未佩戴头盔数量多的路口进行重点整治；又可以实现对路边未佩戴头盔的违章电动车辆进行车牌识别。若未检测到车牌，则可以判定此电动车属于无牌驾驶。若检测到头盔佩戴的数量大于骑行人数，则可以判定该电动车超载。

最后，实验结果表明，本文研究能在一定程度上辅助交警办公，减小电动车车辆交通事故的发生率，提高电动车“绿色出行，安全守护”的头盔佩戴意识。

Violations of electric vehicle drivers can easily lead to collisions between electric vehicles and moving motor vehicles, threatening the lives of cyclists and accompanying personnel who do not wear helmets or do not wear helmets correctly. Therefore, it is of great theoretical value and practical significance to study the helmet detection and license plate recognition methods of electric vehicle riders. However, the current research objects of artificial intelligence and computer vision in the field of intelligent transportation are mainly automobiles. There is a lack of detection and identification of electric vehicle violations and electric vehicle license plates as a whole. Therefore, this paper conducts research on electric vehicles, and the specific work is as follows:
(1) Aiming at the problems of low detection accuracy, poor robustness, and imperfect related systems of electric vehicle small target detection, this paper proposes an improved YOLOv5s algorithm. The improved algorithm uses self-built electric vehicle helmet and license plate data sets, and introduces attention modules CBAM and CA, uses NMS-DIoU to improve non-maximum value suppression NMS, increases multi-scale feature fusion, and realizes automatic detection of electric vehicle helmet wearing and license plate effects. The results show that the overall precision of the improved YOLOv5s is 4.3% higher than that of YOLOv5s, the recall rate is 2.7% higher, and the average precision (IoU=0.5) is 3.6% higher. In the detection of electric vehicle helmets, it can better meet the accuracy requirements.
(2) Aiming at the low accuracy of electric vehicle license plate detection and positioning, and the difficulty in double-line license plate recognition. In this paper, the improved YOLOv5s is used to perform model training on the electric vehicle dataset to improve the accuracy of license plate detection. And through the use of digital image processing technology to perform Hough transform correction on the detected electric vehicle license plate, and at the same time combine the constructed CRNN optical character license plate text recognition model to recognize the electric vehicle license plate text, and the recognition accuracy can reach 91.3%. In the overall automatic detection and recognition process of electric vehicles and license plates, the single frame processing time is about 109 ms.
(3) Through the analysis of actual needs, an automatic detection system for electric vehicle violations is designed. The system realizes various modules including riding an electric vehicle, correctly wearing a helmet, incorrectly wearing a helmet, detection and recognition of electric vehicle license plates, statistical counting, inter-frame delay, picture detection, and video detection. The detection system can realize the detection and statistics of electric vehicle violators who do not wear safety helmets under complex traffic flow, and help law enforcement officers focus on rectifying intersections with large electric vehicle traffic and a large number of road intersections without helmets; License plate recognition of illegal electric vehicles without helmets. If no license plate is detected, it can be determined that the electric vehicle is driven without a license. If it is detected that the number of helmets worn is greater than the number of riders, it can be determined that the electric vehicle is overloaded.
Finally, The experimental results indicate the research in this paper can assist the traffic police to a certain extent, reduce the incidence of electric vehicle traffic accidents, and improve the awareness of wearing helmets for "green travel, safe guarding" of electric vehicles.

[1]苏协.智能微出行绿色双循环——2021中国（无锡）电动车产业发展大会成功举办[J].中国自行车,2021(04):46-47.

[2]Bouhsissin, Soukaina, Sael, et al. Benabbou.Driver Behavior Classification: A Systematic Literature Review[J]. IEEE Access,2013,11:14128-14153.

[3]Jinsheng Xiao, Haowen Guo, Jian Zhou, et al.Tiny object detection with context enhancement and feature purification[J]. Expert Systems with Applications, 2023,211:0957-4174.

[4]Cypto J, Karthikeyan P. Automatic Detection System of Speed Violations in a Traffic Based on Deep Learning Technique[J]. JOURNAL OF INTELLIGENT & FUZZY SYSTEMS, 2022,43(5):6591- 6606.

[5]洪奇峰,施伟斌,吴迪,罗力源.深度卷积神经网络模型发展综述[J].软件导刊,2020,19(04):84-88.

[6]S. Ren, K. He, R. Girshick, et al, Faster r-cnn: towards real-time object detection with region proposal networks[J]. Neural Information Processing Systems, 2015,28.

[7]He K, Gkioxari G, Dollar P, et al. Mask R-CNN[C]// Proceedings of the 2017 IEEE International Conference on Computer Vision.2017:2980-2988.0,19(04):84-88.

[8]Redmon J, Divvala S, Girshick R, et al. You only look once:unified real-time object detection[C]// Proceedings of the 2016 IEEE Conference on Computer Vision and Pattern Recognition. 2016:779-788.

[9]Lin T Y, Goyal P, Girshick R,et al. Focal loss for dense object detection[C]// Proceedings of the 2017 IEEE International Conference on Computer Vision. 2017:2999-3007.

[10]Liu W, Anguelov D, Erhan D,et al. SSD: single shot multiBox detector [C]// Proceedings of the 2014 European Conference on Computer Vision,LNCS 9905. Cham: Springer,2016:21-37.

[11]Li Y, Xiong X, Xin W, et al. MobileNetV3-CenterNet:A Target Recognition Method for Avoiding Missed Detection Effectively Based on a Lightweight Network[J].Journal of Beijing Institute of Technology,2023,32(01):82-94.

[12]TIAN Z, SHEN C H, CHEN H, et al. FCOS: fully convolutional one-stage object detection[C]// Proceedings of the 2019 IEEE/CVF International Conference on Computer Vision. 2019:9626-9635.

[13]Gao J, Chen Y, Wei Y, et al. Detection of Specific Building in Remote Sensing Images Using a Novel YOLO-S-CIOU Model[J]. Gas Station Identification. 2021, 21: 1375.

[14]Wang, Chien-Yao, Alexey Bochkovskiy, et al. YOLOv7: Trainable bag-of-freebies sets new state-of-the-art for real-time object detectors[J]. arXiv ,2022, abs/2207.02696.

[15]赵睿,刘辉,刘沛霖,雷音,李达.基于改进YOLOv5s的安全帽检测算法[J/OL].北京航空航天大学学报:1-16[2023-04-05].DOI:10.13700/j.bh.1001-5965.2021.0595.

[16]BHagat S,Contractor D,Sharma S,et al. Cascade classifier based helmet detection using OpenCV in image processing[C]// Proceedings of the 2016 National Conference on Recent Trends in Computer and Communication Technology.[2021-02-28].

[17]Silva R, Aires K, Santos T, et al. Automatic detection of motorcyclists without helmet[C]// Proceedings of the XXXIX Latin American Computing Conference. IEEE,2013:1-7.

[18]Yogameena B,Menaka K,Perumaal S S. Deep learning based helmet wear analysis of a motorcycle rider for intelligent surveillance system[J]. IET Intelligent Transport Systems，2019，13（7）:1190-1198.

[19]Lim J-J, Kim D-W, Hong W-H, et al. Application of Convolutional Neural Network (CNN) to Recognize Ship Structures. Sensors. 2022,22(10):3824.

[20]Vishnu C, Singh D, Mohan C K, et al. Detection of motorcyclists without helmet in videos using convolutional neural network[C]//Proceedings of the International Joint Conference on Neural Networks.IEEE,2017:3036-3041.

[21]Shine L,Jiji C V. Automated detection of helmet on motorcyclists from traffic surveillance videos: a comparative analysis using hand-crafted features and CNN[J]. Multimedia Tools Applications，2020，79（19/20）:14179-14199.

[22]ChairaT A, Dailey M N, Limsoonthrakul S, et al. Low cost,high performance automatic motorcycle helmet violation detection[C]// Proceedings of the 2020 IEEE Winter Conference on Applications of Computer Vision. IEEE,2020:3549-3557.

[23]Singh D, Vishnu C, Mohan C K. Real-time detection of motorcyclist without helmet using cascade of CNNs on edge-device[C]// Proceedings of the IEEE 23rd International Conference on Intelligent Transportation Systems. IEEE,2020:1-8.

[24]DAsgupta M, Bandyopadhyay O, Chatterji S. Automated helmet detection for multiple motorcycle riders using CNN [C]// Proceedings of the 2019 IEEE Conference on Information and Communication Technology. IEEE,2019:1-4.

[25]徐振博. 基于视觉的车牌与车辆的检测、识别与追踪技术研究[D].合肥:中国科学技术大学,2021.

[26]Tan C, Cao J. An Algorithm for License Plate Location Based on Color and Texture[C]//International Conference on Intelligent Human-Machine Systems and Cybernetics. IEEE Computer Society,2013:356-358.

[27]Björklund T, Fiandrotti A, Annarumma M, et al. Robust license plate recognition using neural networks trained on synthetic images[J]. Pattern Recognition, 2019, 93: 134-146.

[28]R. Girshick. Fast R-CNN[C]// Proceedings of the IEEE International Conference on Computer Vision. 2015:1440–1448.

[29]Fan Q, Zhuo W, Tang C K, et al. Few-shot object detection with attention-RPN and multi-relation detector[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2020: 4013-4022.

[30]T. Nazir, A. Irtaza, J. Rashid, et al.Diabetic retinopathy lesions detection using faster-RCNN from retinal images[C]//Proceedings of the 2020 First International Conference of Smart Systems and Emerging Technologies (SMARTTECH).2020: 38-42.

[31]Q.Huang, Z. Cai and T. Lan. A New Approach for Character Recognition of Multi-Style Vehicle License Plates[J]. IEEE Transactions on Multimedia,2021,23:3768-3777.

[32]Lee Y, Lee J, Ahn H, et al. Snider: Single noisy image denoising and rectification for improving license plate recognition[C]//Proceedings of the IEEE/CVF International Conference on Computer Vision Workshops. 2019:67-72.

[33]Bulan O, Kozitsky V, Ramesh P, et al. Segmentation-and annotation-free license plate recognition with deep localization and failure identification[J].IEEE Transactions on Intelligent Transportation Systems, 2017, 18(9): 2351-2363.

[34]Lan W, Dang J, Wang Y, et al. Pedestrian detection based on YOLO network model[C]//2018 IEEE international conference on mechatronics and automation (ICMA). IEEE, 2018: 1547-1551.

[35]刘艳菊,伊鑫海,李炎阁,张惠玉,刘彦忠.深度学习在场景文字识别技术中的应用综述[J].计算机工程与应用,2022,58(04):52-63.

[36]陈巧巧.电动自行车违章自动检测系统的研究与实现[D].海口:海南大学,2020.

[37]Ribeiro V, Greati V, Bezerra A , et al. Brazilian Mercosur License Plate Detection: a Deep Learning Approach Relying on Synthetic Imagery[C]// 2019 IX Brazilian Symposium on Computing Systems Engineering (SBESC). IEEE, 2019.

[38]Kao C C, Wang W, Sun M, et al. R-CRNN: Region-based convolutional recurrent neural network for audio event detection[J]. arXiv ,2018,1808.06627:34-45.

[39]Tong G, Li Y, Gao H, et al. MA-CRNN: a multi-scale attention CRNN for Chinese text line recognition in natural scenes[J]. Document Analysis and Recognition, 2020, 23(12).

[40]李永上,马荣贵,张美月.改进YOLOv5s+DeepSORT的监控视频车流量统计[J].计算机工程与应用,2022,58(05):271-279.

[41]He K M, Zhang X Y, Ren S Q, et al. Spatial pyramid pooling in deep convolutional networks for visual recognition[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence.2015,37(9):1904-1916.

[42]Lin T Y, Dollar P, Girshick R, et al. Feature pyramid networks for object detection[C]// Proceedings of the 2017 IEEE Conference on Computer Vision and Pattern Recognition.IEEE,2017:936-944.

[43]Liu S, Qi L, Qin H F, et al. Path aggregation network for instance segmentation[C]// Proceedings of the 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. IEEE,2018:8759-8768.

[44]Avci C, Tekinerdogan B, Catal C. Analyzing the performance of long short-term memory architectures for malware detection models. CONCURRENCY AND COMPUTATION PRACTICE & EXPERIENCE. 2023, 35(6).

[45]Weerakody, P.B, K.W. et al. Cyclic Gate Recurrent Neural Networks for Time Series Data with Missing Values[EB/OL]. Neural Processing Letters, 2022-07-30.

[46]Tao Zhou, XinYu Ye, HuiLing Lu, et al.Dense Convolutional Network and Its Application in Medical Image Analysis. BioMed Research International, 2022,vol. 2022, 2384830.

[47]J. S. Kang and K. Chung. STAug:Copy-Paste Based Image Augmentation Technique Using Salient Target[J].IEEE Access, 2022,10:123605-123613.

[48]Sun X , Hao H ,Liu Y , et al. Research on the Application of YOLOv4 in Power Inspection[J]. IOP Conference Series Earth and Environmental Science, 2021, 693(1):012038.

[49]Lotter W, Diab A R, Haslam B , et al. Robust breast cancer detection in mammography and digital breast tomosynthesis using an annotation-efficient deep learning approach[J]. Nature Medicine, 2021，27(2):244–249.

[50]Xu D, Wang L, Li F. Research summary of typical target detection algorithms for deep learning[J]. Computer Engineering and Applications, 2021, 57(8):10-215.

[51]Al-Amri S S, Kalyankai N V, Khamitkar S D. Linear and non-linear contrast enhancement image[J]. Amri, 2010, 10(2): 139-143.

[52]Bocketein I M. Color equalization method and its application to color image processing[J]. Journal of the Optical Society of America A, 1986, 3(5): 735-737.

[53]刘迪, 贾金露, 赵玉卿, 钱育蓉. 基于深度学习的图像去噪方法研究综述[J]. 计算机工程与应用, 2021, 57(07): 1-13.

[54]邹梓吟,盖绍彦,达飞鹏,李昱.基于注意力机制的遮挡行人检测算法[J].光学学报,2021,41(15):157-165.

[55]Woo S, Park J, Lee J Y, et al. CBAM: convolutional block attention module[J]. Springer,Charm,2018:3-19.

[56]Hou Q B, Zhou D Q, Feng J S. Coordinate attention for efficient mobile network design[C]// Proceedings of the 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition.IEEE,2021:13708-13717.

[57]Wu W K,Zhang Y,Wang D,et al. SK-Net: deep learning on point cloud via end-to-end discovery of spatial keypoints[C]// Proceedings of the 34th AAAI Conference on Artificial Intelligence. Palo Alto,CA:AAAI Press,2020:6422-6429.

[58]Zhang Z H, Wang P, Liu W, et al. Distance-IoU loss:faster and better learning for bounding box regression[C]//Proceedings of the 34th AAAI Conference on Artificial Intelligence. Palo Alto,CA:AAAI Press,2020:12993-13000.

[59]Li S S, Li Y J, Li Y, et al. YOLO-FIRI:improved YOLOv5 for infrared image object detection[J]. IEEE Access,2021,9:141861-141875.

[60]樊缤,李智,高健.基于多尺度知识学习的深度鲁棒水印算法[J/OL]. 计算机应用,2022,42(10):3102-3110.