摘要
针对多目标跟踪问题,提出了一种基于目标检测和联合概率数据关联的方法。将目标检测的结果作为系统的观测值,采用联合概率数据关联方法对它们在时间上做数据关联,最后采用卡尔曼滤波对关联好的观测值和状态值做滤波、跟踪。
This paper presents a multi-target tracking method based on object detection and joint probability data association.It takes the results of object detections as the observations and uses joint probability data association method to make data associations between observations and system states.At last the classic Kalman filter is adopted to filter the system states and give the final tracking results.
出处
《工业控制计算机》
2018年第3期17-19,22,共4页
Industrial Control Computer
关键词
目标检测
多目标跟踪
联合概率数据关联
YOLO
object detection
multi-target tracking
joint probability data association
YOLO
