基于人机交互的免锚检测和跟踪系统设计  

Anchor-Free Detection and Tracking System Design Based on Human-Computer Interaction

作　　者：吴超[1] 吴绍斌[1] 李子睿 孙冬 WU Chao;WU Shaobin;LI Zirui;SUN Dong(School of Mechanical Engineering,Beijing Institute of Technology,Beijing 100081,China)

机构地区：[1]北京理工大学机械与车辆学院,北京100081

出　　处：《兵工学报》2022年第10期2565-2575,共11页Acta Armamentarii

摘　　要：为提高越野环境中目标检测和跟踪的准确率和效率,提出一种基于人机交互的免锚检测和跟踪系统。该系统由检测系统、指挥系统和目标跟踪系统组成。检测系统,在基于点的点云特征提取框架的基础上,设计一种免锚的目标检测网络结构;指挥系统通过相机实时获取环境态势信息,人机交互地在检测网络输出的目标序列中选择跟踪目标;跟踪系统利用检测网络输出的目标序列的外观模型和指挥系统下发的跟踪目标外观模型进行匹配来确定跟踪目标,再基于卡尔曼滤波算法进行目标运动估计。基于越野场景的实车数据进行了验证。验证结果表明:基于人机交互的免锚检测和跟踪算法在不增加算法时间的同时实现了超91%的准确率,能够满足无人驾驶车辆在越野场景的使用要求。To improve the accuracy and efficiency of object detection and tracking in the off-road environment,an anchor-free object detection and tracking system based on human-computer interaction is proposed.It consists of detection system,command system and object tracking system.For detection system,on the basis of point-based point cloud feature extraction framework,an anchor-free object detection network structure is designed.The command system can display information on real-time environmental situation and select targets interactively in the objects sequence output by the detection network.For tracking system,the appearance model in the objects sequence output by the detection system is used to match with that of the tracked target issued by the comand system to determine the target being tracked.Then the Kalman filter algorithm is adopted for objects motion estimation.The verification analysis is done based on real vehicle data in the off-road environment.The results show that the proposed system achieved an accuracy rate of over 91%without increasing the time taken by the algorithm,which can meet the requirements of autonomous vehicles in off-road scenarios.

关 键 词：3D目标检测 免锚 人机交互 外观模型 运动估计 

分 类 号：TP242.6[自动化与计算机技术—检测技术与自动化装置]