[智能感知]基于YOLO_v2模型的车辆实时检测

摘要/Abstract

摘要：

为了解决传统车辆检测实时性差和摄像头获取信息单一的问题，提出了一种基于改进YOLO_v2模型的车辆实时检测算法。基于YOLO_v2网络结构建立车辆检测模型，证明了YOLO_v2算法在车辆检测方面准确率高、实时性好。对YOLO_v2算法进行改进，使改进后的算法能对采集到的车载视频信息进行多维度判断：判断图片中是否有车辆及车辆在图片中的位置，判断被检测车辆与摄像头的相对方位及运动趋势，判断被检测车辆对自身车辆的危险程度。实验结果表明，改进后的模型在车载视频上取得了良好的检测效果，解决了车载视频中车辆检测实时性低的问题，并将传统基于视觉的车辆检测从单一维度检测扩展到了多维度检测。

关键词: YOLO_v2模型, 车辆检测, 车载视频, 实时, 多维度的

Abstract: In order to solve the problems of poor real-time detection and single acquisition information from vehicle-mounted camera， a real-time vehicle detection algorithm was proposed based on improved YOLO_v2 model. Based on the YOLO_v2 network structure， a vehicle detection model was established， which proved that the YOLO_v2 algorithm had high accuracy and good real-time performance in vehicle detections. And the YOLO_v2 algorithm was improved so that the improved one might perform multi-dimensional judgments on the vehicle-mounted video informations： judging whether there was a vehicle and the vehicle positions in the pictures， judging the relative positions to camera and the movement trend of the detected vehicles， judging the danger degree of the detected vehicles to the own vehicle. The experimental results show that the improved model achieves good detection effectiveness on vehicle-mounted video， solves the problems of low real-time vehicle detections in vehicle-mounted video， extends the traditional vision-based vehicle detections from single dimensional detections to multi-dimensional detections.

Key words: YOLO_v2 model, vehicle detection, vehicle video, real-time

中图分类号:

TP391.4

黎洲, 黄妙华, . [智能感知]基于YOLO_v2模型的车辆实时检测[J]. 中国机械工程.

LI Zhou, HUANG Miaohua, . Vehicle Detections Based on YOLO_v2 in Real-time[J]. China Mechanical Engineering.

参考文献

[1]KACHACH R， CANAS J M. Hybrid Three-dimensional and Support Vector Machine Approach for Automatic Vehicle Tracking and Classification Using a Single Camera [J］.Journal of Electronic Imaging， 2016， 25（3）：24.
[2]FELZENSZWALB P， MCALLESTER D， RAMANAN D. A Discriminatively Trained， Multiscale， Deformable Part Model[C］//Computer Vision and Pattern Recognition. Boston： IEEE， 2008：1-8.
[3]FELZENSZWALB P， GIRSHICK R B， MCALLESTER D， et al. Object Detection with Discriminatively Trained Part-based Models [J］.IEEE Transactions on Pattern Analysis and Machine Intelligence， 2010， 32（9）：1627-1645.
[4]KRIZHEVSKY A， SUTSKEVER I， HINTON G E. ImageNet Classification with Deep Convolutional Neural Networks[C］//International Conference on Neural Information Processing Systems. USA： Curran Associates， 2012：1097-1105.
[5]GIRSHICK R， DONAHUE J， DARRELL T， et al. Rich Feature Hierarchies for Accurate Object Detection and Semantic Segmentation[C］//Computer Vision and Pattern Recognition. Boston： IEEE， 2013：580-587.
[6]UIJLINGS J R， van de SANDE K E， GEVERS T， et al. Selective Search for Object Recognition[C］//International Journal of Computer Vision. Hawaii： IJCV Press， 2013： 115-117.
[7]HE K， ZHANG X， REN S， 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.
[8]GIRSHICK R. Fast R-CNN [C］//IEEE International Conference on Computer Vision. Santiago： IEEE， 2015：10-15.
[9]REN S， HE K， GIRSHICK R， et al. Faster R-CNN： Towards Real-time Object Detection with Region Proposal Networks[J］. IEEE Transactions on Pattern Analysis and Machine Intelligence， 2017， 39（6）：1137-1149.
[10]DAI Jifeng， LI Yi， HE Kaiming， et al. R-FCN： Object Detection via Region-based Fully Convolutional Networks [EB/OL］. Ithaca， New York： Cornell University Library， 2016 [2017-11-01］. https：//arxiv.org/abs/1605.06409.
[11]REDMON J， DIVVALA S， GIRSHICK R， et al. You Only Look Once： Unified， Real-time Object Detection[C］// Computer Vision and Pattern Recognition. Boston： IEEE， 2016：779-788.
[12]LIU W，ANGUELOV D，ERHAN D，et al. SSD： Single Shot MultiBox Detector [EB/OL］. Ithaca， New York： Cornell University Library， 2015 [2017-11-1］. https：//arxiv.org/abs/1512.02325.
[13]REDMON J， FARHADI A. YOLO9000： Better， Faster， Stronger [EB/OL］. Ithaca， New York： Cornell University Library， 2016 [2017-11-01］. https：//arxiv.org/abs/1612.08242.
[14]ZHANG J， HUANG M， JIN X， et al. A Real-time Chinese Traffic Sign Detection Algorithm Based on Modified YOLO_v2 [J］. Algorithms， 2017， 10（4）：127-139.
[15]JO K U，IM J H， KIM J，et al. A Real-time Multi-class Multi-object Tracker Using YOLO_v2[C］//IEEE International Conference on Signal and Image Processing Applications. Beijing：IEEE， 2017：507-511.
[16]SEO C J. The Study of Car Detection on the Highway Using YOLO_v2 and UAVs [J］. Transaction of the Korean Institute of Electrical Engineers P， 2018， 9（4）：67-73.
[17]魏湧明，全吉成，侯宇青阳.基于YOLO_v2的无人机航拍图像定位研究[J］.激光与光电子学进展，2017， 54（11）：95-104.

WEI Yongming， QUAN Jicheng， Hou Yuqing- yang. Research on Aerial Image Positioning of UAV Based on YOLO_v2 [J］.Progress in Laser and Optoelectronics， 2017， 54（11）：95-104.
[18]王宇宁，庞智恒，袁德明.基于YOLO算法的车辆实时检测[J］.武汉理工大学学报，2016，38（10）：41-46.

WANG Yuning， PANG Zhiheng， YUAN Deming. Vehicle Detection Based on YOLO in Real Time [J］. Journal of Wuhan University of Technology， 2016， 38（10）：41-46.
[19]NEUBECK A， GOOL L V. Efficient Non-maximum Suppression[C］//18th International Conference on Pattern Recognition. Hong Kong， 2006：850-855.