Traffic Flow Statistics Method and Practice Based on Roadside Video with YOLO_V3
赖见辉1,2, 罗甜甜2, 王扬2, 陈艳艳2
1. Opening Funding Supported by the Platform of Transport Technology Thinktank, Research Institute of Highway, Ministry of Transport, Beijing 100088, China;
2. College of Metropolitan Transportation, Beijing University of Technology, Beijing 100124, China
Traffic Flow Statistics Method and Practice Based on Roadside Video with YOLO_V3
LAI Jian-hui1,2, LUO Tian-tian2, WANG Yang2, CHEN Yan-yan2
1. Opening Funding Supported by the Platform of Transport Technology Thinktank, Research Institute of Highway, Ministry of Transport, Beijing 100088, China;
2. College of Metropolitan Transportation, Beijing University of Technology, Beijing 100124, China
摘要Temporary traffic flow observation plays a pivotal role in transportation planning, consultation, and decision-making, and it supplements fixed long-term traffic observation data. Although substantial automated technology has been used for traffic observation, such technology is limited by the temporariness and uncertainty of observation points; moreover, these methods are difficult to use in temporary observations. In this study, the YOLO_V3 algorithm, which is based on deep learning, is used for vehicle detection based on roadside videos of temporary observations. Moreover, a secondary detection framework based on vehicle detection and traffic counting regions is proposed. Then, a traffic counting pattern with Kalman filter, Hungarian allocation, and perspective projection transformation is established. In addition, by collecting multiple sets of actual video data, the effectiveness of the method under different conditions is analyzed in terms of three indicators:camera intersection angle with the road, erection height, and road traffic density. Results show that at a camera height of 3 m and a roadside angle of 30°, the accuracy is approximately 95%. However, traffic flow accuracy drops to approximately 90% when vehicles are blocked by large buses and trucks during detection. In this study, the algorithm is tested for execution efficiency using 1080p video streams on a Windows 10 x 64 operating system, a 2080Ti graphics card, a 64 GB RAM, and an i7-7820X CPU. Results show that camera installation angle and height have no considerable effect on operation efficiency. Under low-density traffic, the frames per second (FPS) value is approximately 44; under high-density traffic, the FPS value drops to approximately 33, indicating that the method still has high execution efficiency and can be used for real-time video traffic counting.
Abstract：Temporary traffic flow observation plays a pivotal role in transportation planning, consultation, and decision-making, and it supplements fixed long-term traffic observation data. Although substantial automated technology has been used for traffic observation, such technology is limited by the temporariness and uncertainty of observation points; moreover, these methods are difficult to use in temporary observations. In this study, the YOLO_V3 algorithm, which is based on deep learning, is used for vehicle detection based on roadside videos of temporary observations. Moreover, a secondary detection framework based on vehicle detection and traffic counting regions is proposed. Then, a traffic counting pattern with Kalman filter, Hungarian allocation, and perspective projection transformation is established. In addition, by collecting multiple sets of actual video data, the effectiveness of the method under different conditions is analyzed in terms of three indicators:camera intersection angle with the road, erection height, and road traffic density. Results show that at a camera height of 3 m and a roadside angle of 30°, the accuracy is approximately 95%. However, traffic flow accuracy drops to approximately 90% when vehicles are blocked by large buses and trucks during detection. In this study, the algorithm is tested for execution efficiency using 1080p video streams on a Windows 10 x 64 operating system, a 2080Ti graphics card, a 64 GB RAM, and an i7-7820X CPU. Results show that camera installation angle and height have no considerable effect on operation efficiency. Under low-density traffic, the frames per second (FPS) value is approximately 44; under high-density traffic, the FPS value drops to approximately 33, indicating that the method still has high execution efficiency and can be used for real-time video traffic counting.
基金资助:Supported by the Beijing Municipal Science and Technology Project (No.Z181100003918011)
通讯作者:
LAI Jian-hui
E-mail: laijianhui@bjut.edu.cn
引用本文:
赖见辉, 罗甜甜, 王扬, 陈艳艳. Traffic Flow Statistics Method and Practice Based on Roadside Video with YOLO_V3[J]. Journal of Highway and Transportation Research and Development, 2021, 15(1): 89-97.
LAI Jian-hui, LUO Tian-tian, WANG Yang, CHEN Yan-yan. Traffic Flow Statistics Method and Practice Based on Roadside Video with YOLO_V3. Journal of Highway and Transportation Research and Development, 2021, 15(1): 89-97.
[1] JIANG Gui-yan, GUO Hai-feng, WU Chao-teng. Identification Method of Urban Road Traffic Conditions Based on Inductive Coil Data[J]. Journal of Jilin University:Engineering and Technology Edition. 2008, 38(S1):37-42. (in Chinese)
[2] IWASAKI Y, MISUMI M, NAKAMIYA T. Robust Vehicle Detection under Various Environmental Conditions Using an Infrared Thermal Camera and Its Application to Road Traffic Flow Monitoring[J]. Sensors. 2013, 13(6):7756-7773.
[3] JO Y, JUNG I. Analysis of Vehicle Detection with WSN-Based Ultrasonic Sensors[J]. Sensors. 2014, 14(8):14050-14069.
[4] WANG Xiang, XIAO Jian-li. Overview of Video-based Traffic Parameters Detection[J]. Journal of University of Shanghai for Science and Technology. 2016, 38(5):479-486. (in Chinese)
[5] KACHROO P, SHLAYAN N, PAZ A, et al. Model-Based Methodology for Validation of Traffic Flow Detectors by Minimizing Human Bias in Video Data Processing[J]. IEEE Transactions on Intelligent Transportation Systems. 2015, 16(4):1851-1860.
[6] ZHANG Run-chu. Research on Video-based Traffic Flow Parameters Extraction Method and System Realization[D]. Guangzhou:South China University of Technology, 2015. (in Chinese)
[7] LI Dong. Traffic Statistical Algorithm Research Based on Video Image[D]. Dalian:Dalian Maritime University, 2016. (in Chinese)
[8] LIU Chang. Research of Vehicle Flow Statistics Based on Video Image[D]. Dalian:Dalian Jiaotong University, 2016. (in Chinese)
[9] HU Yun-lu. Study on the Traffic Flow and Vehicle Speed Detection System Based on Video[D]. Xi'an:Chang'an University, 2017. (in Chinese)
[10] DAI Feng-zhi, WEI Bao-chang, OUYANG Yu-xing, et al. Survey of Research Progress of Video Tracking Based on Deep Learning[J]. Computer Engineering and Applications. 2019, 55(10):16-29. (in Chinese)
[11] WU Yu-lu, ZHANG De-xian. A Survey of Target Detection Algorithms Based on Deep Learning[J]. China Computer & Communication. 2019(12):46-48. (in Chinese)
[12] LIU Yi-ming. Research and Application Review of Deep Learning[J]. Journal of Green Science and Technology. 2019(11):281-283. (in Chinese)
[13] CHU Xiang-yu. Research on Traffic Video Detection and Vehicle Classification Based on Deep Learning[D]. Harbin:Harbin Institute of Technology, 2017. (in Chinese)
[14] NIU Jia-jun. Design and Implementation of Traffic Flow Tracking Statistics System Based on Video Processing[D]. Chengdu:University of Electronic Science and Technology of China, 2018. (in Chinese)
[15] REDMON J, DIVVALA S, GIRSHICK R, et al. You Only Look Once:Unified, Real-Time Object Detection[C]//IEEE Conference on Computer Vision and Pattern Recognition. Las Vegas:IEEE, 2016, 779-788.
[16] SHI Hui, CHEN Xian-qiao, YANG Ying. Safety Helmet Wearing Detection Method of Improved YOLO v3[J]. Computer Engineering and Applications. 2019, 55(11):213-220. (in Chinese)
[17] WANG Chao, FU Zi-ang. Traffic Sign Detection Based on YOLO v2 Model[J]. Journal of Computer Applications. 2018, 38(S2):276-278. (in Chinese)
[18] GUO Jin-xiang, LIU Li-bo, XU Feng, et al. Airport Scene Aircraft Detection Method Based on YOLO v3[J]. Laser & Optoelectronics Progress, 2019,56(19):103-111. (in Chinese)
[19] WANG Fu-jian, ZHANG Jun, LU Guo-quan, et al. YOLO-based Vehicle Information Detection and Tracking System[J]. Industrial Control Computer. 2018, 31(7):89-91. (in Chinese)
[20] WANG Yu-ning, PANG Zhi-heng, YUAN De-ming. Vehicle Detection Based on YOLO in Real Time[J]. Journal of Wuhan University of Technology. 2016, 38(10):41-46. (in Chinese)
[21] ZHANG Jia-chen, CHEN Qing-kui. Road Vehicle Congestion Analysis Model Based on YOLO[J]. Journal of Computer Applications. 2019, 39(1):93-97. (in Chinese)
[22] LI Jun-li, YIN Kuan, CHU Cheng-xi, et al. Review of Video Target Tracking Technology[J]. Journal of Yanshan University. 2019, 43(3):251-262. (in Chinese)
[23] CHAI Tian-feng. Image Object Tracking Algorithm for Aircraft Oriented Platform[D]. Mianyang:China Academy of Engineering Physics, 2017. (in Chinese)
[24] DRUMMOND T, CIPOLLA R. Real-time Visual Tracking of Complex Structures[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence. 2002, 24(7):932-946.
[25] LIU R, LU Y. Infrared Target Tracking in Multiple Feature Pseudo-color Image with Kernel Density Estimation[J]. Infrared Physics & Technology. 2012, 55(6):505-512.
[26] KALMAN E R. A New Approach to Linear Filtering and Prediction Problems[J]. Journal of Basic Engineering Transactions. 1960, 82(1):35-45.
[27] GORDON N J, SALMOND D J, SMITH A. Novel-APPROACH to Nonlinear Non-Gausslan Bayeslan State Estimation[J]. IEE Proceedings F:Radar and Signal Processing. 1993, 140(2):107-113.
2021, Vol. 15 
