基于YOLOv3和DeepSort的车流量检测

针对传统多目标跟踪算法的检测跟踪精度低、鲁棒性差的缺点,基于经典的Tracking-By-Detection模式,提出一种基于YOLOv3和DeepSort的车流量检测方法,实现了车辆视频监控端到端的车流量视频的实时监测与跟踪计数。采用深度学习YOLOv3算法检测视频车辆目标,然后利用深度学习DeepSort算法对检测到的车辆进行实时跟踪计数。实验结果表明该方法应对快速移动的车辆和环境光照的影响时,对车流量的检测效果良好,平均精度达到94.7%,端到端的算法可行且有效,适用于对车辆视频的批处理。

Vehicle Flow Detection Based on YOLOv3 and DeepSort

In view of the shortcomings of the traditional multi-target tracking algorithm, such as low detection accuracy and poor robustness, according to the classic Tracking-By-Detection mode, a vehicle flow detection method based on YOLOv3 and DeepSort is proposed, which realizes the real-time monitoring and tracking of the end-to-end vehicle flow video in vehicle video monitoring. The video vehicle target is detected by deep learning YOLOv3 algorithm, and then the detected vehicle is tracked in real time by deep learning DeepSort algorithm. The experimental results show that the method has good detection effect on traffic flow when dealing with the influence of fast moving vehicles and ambient light, and the average accuracy is up to 94.7%. The end-to-end algorithm is feasible and effective, which is suitable for video batch processing.