基于人机交互的免锚检测和跟踪系统设计

doi:10.12382/bgxb.2021.0682

摘要/Abstract

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。

关键词: 3D目标检测, 免锚, 人机交互, 外观模型, 运动估计

Abstract: 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.

Key words: 3Dobjectdetection, anchor-free, human-computerinteraction, appearancemodel, motionestimation

中图分类号:

TP242.6

吴超, 吴绍斌, 李子睿, 孙冬. 基于人机交互的免锚检测和跟踪系统设计[J]. 兵工学报, 2022, 43(10): 2565-2575.

WU Chao, WU Shaobin, LI Zirui, SUN Dong. Anchor-Free Detection and Tracking System Design Based on Human-Computer Interaction[J]. Acta Armamentarii, 2022, 43(10): 2565-2575.

参考文献

［1］谢苏明.无人化智能化装备技术发展及其影响分析[J].现代军事，2017(3):51-56．
XIE S M.The development of unmanned intelligent equipment technology and its influence analysis[J].Modern Military，2017(3): 51-56.(in Chinese)
[2］陈慧岩，张玉.军用地面无人机动平台技术发展综述[J].兵工学报，2014，35(10):1696-1706.
CHEN H Y，ZHANG Y.Summary of the development of military ground UAV mobile platformtechnology[J].Acta Armamentarii，2014，35(10):1696-1706.(in Chinese)
[3］ NEGRE A，RUMMELHARD L，LAUGIER C.Hybrid sampling Bayesian occupancy filter[C]∥Proceedings of 2014 IEEE Intelligent Vehicles Symposium Proceedings.Dearborn，MI，US:IEEE，2014:1307-1312.
[4］ DANESCU R，ONIGA F，NEDEVSCHI S.Modeling and tracking the driving environment with a particle based occupancy grid[J].IEEE Transactions on Intelligent Transportation Systems，2011，12(4): 1331-1342.
[5］ YANG Y，YAN G，ZHU H，et al. Moving object detection under dynamic background in 3D range data[C]∥Proceedings of 2014 IEEE Intelligent Vehicles Symposium Proceedings. Dearborn，MI，US:IEEE，2014: 394-399.
[6］ LI Z，DU Y，ZHU M.et al.A survey of 3D object detection algorithms for intelligent vehicles development[J].Artif Life Robotics，2022，27：115-122.
[7］ YIN Z，ONCEL T.Voxel net:end-to-end learning for point cloud based 3D object detection[C]∥Proceedings of 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Salt Lake City，UT，US:IEEE，2018:4490-4499.
[8］ LANG A H，VORA S，CAESAR H，et al. Pointpillars:fast encoders for object detection from point clouds[C]∥Proceedings of 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Long Beach，CA，US:IEEE，2019:12689-12697.
[9］ YIN T T，ZHOU X，P. Krhenbühl. Center-based 3D object detection and tracking[C]∥Proceedings of 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition.Virtual:IEEE，2021：11779-11788，
[10］ WANG Q，CHEN J，DENG J Q，et al. 3D-CenterNet: 3D object detection network for point clouds with center estimation priority[J].Pattern Recognition，2021，115:107884
[11］ QI C R，SU H，MO K,et al.Pointnet: deep learning on point sets for 3d classification and segmentation[C]∥Proceedings of 2017 IEEE Conference on Computer Vision and Pattern Recognition. Honolulu，HI，US: IEEE，2017:77-85
[12］ QI C R，YI L，SU H，et al. Pointnet++:deep hierarchical feature learning on point sets in a metric space[C]∥Proceedings of Neural Information Processing Systems. CA，US:NIPS，2017: 5099-5108.
[13］ SHI S H，WANG X G，LI H S.Pointrcnn: 3d object proposal generation and detection from point cloud[C]∥Proceedings of 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition.Long Beach，CA，US:IEEE，2019:770-779.
[14］ SHI S S，GUO C X，JIANG L. Point-voxel feature set abstraction for 3D object detection[C]∥Proceedings of 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Seattle，WA，US:IEEE，2020:10526-10535.
[15］ CHEN X Z，MA H M，WAN J，et al.Multi-view 3d object detection network for autonomous driving[C]∥Proceedings of 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Long Beach，CA，US:IEEE，2019: 1907-1915.
[16］ QI C R，LIU W，WU C X，et al. Frustum pointnets for 3d object detection from RGBD data[C]∥Proceedings of 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition.Salt Lake City，UT，US:IEEE，2018:918-927.
[17］ HONG D S，CHEN H H，HSIAO P Y，et al.Cross fusion net Deep 3D object detection based on RGB images and point clouds in autonomous driving[J].Image and Vision Computing，2020，100:103955
[18］张长弓，杨海涛，王晋宇，等.基于深度学习的视觉单目标跟踪综述[J].计算机应用研究，2021，38(10):2888-2895.
ZHANG C G，YANG H T，WANG J Y,et al. Overview of single-target tracking based on deep learning[J].Computer Application Research，2021，38(10): 2888-2895.(in Chinese)
[19］ YANG B，WANG J N，CLARK R. Learning object bounding boxes for 3D instance segmentation on point clouds[C]∥Proceedings of Neural Information Processing Systems. Vancouver，Canada:Neural Information Processing Systems，2019:6737-6746.
[20］ LIN T Y，GOYAL P，GIRSHICK R，et al. Focal loss for dense object detection[J].IEEE Transactions on Pattern Analysis and Machine Intelligence，2018:318-327.
[21］董伟，高凯. 智能化战争呼唤指挥智能化[N].中国国防报，2019-06-26(003).
DONG W，GAO K. Intelligent warfare calls for command intelligence[N].China National Defense News website.2019-06-26(003).(in Chinese)
[22］ WOJKE N，BEWLEY A，PAULUS D.Simple online and real time tracking with a deep association metric[C]∥Proceedings of 2017 IEEE International Conference on Image Processing.Beijing，China:IEEE，2017:3645-3649.
[23］ GEIGER A，LENZ P，URTASUN R. Are we ready for autonomous driving? the KITTI vision benchmark suite[C]∥Proceedings of 2012 IEEE Conference on Computer Vision and Pattern Recognition. Providence，RI，US: IEEE，2012:3354-3361.

[1]	王鑫强，王博，贾巍，先毅，杨宗睿，韩京冶. 基于Wayland图形协议和光纤通道网络的车辆人机交互显示控制技术[J]. 兵工学报, 2022, 43(S1): 46-53.
[2]	乔宇，穆煜，吕韫琦，高学山. 康复助行机器人仿真设计与分析[J]. 兵工学报, 2018, 39(4): 825-832.
[3]	吴浩, 邓宏彬. 一种基于分块灰度投影运动估计的视频稳像方法[J]. 兵工学报, 2013, 34(4): 406-411.
[4]	江燕华，熊光明，姜岩，龚建伟，陈慧岩. 智能车辆视觉里程计算法研究进展[J]. 兵工学报, 2012, 33(2): 214-220.
[5]	徐ー鸣，刘晓利，刘怡昕. 基于分割模板运动预估的相关跟踪算法[J]. 兵工学报, 2009, 30(7): 879-884.
[6]	卓志敏，杨莘元，杨雷. 基于RANSAC + LS算法的红外成像全局运动设计[J]. 兵工学报, 2008, 29(3): 308-312.
[7]	那慧，颜景龙，张树江. 基于Kalman滤波的稳像技术研究[J]. 兵工学报, 2007, 28(2): 175-177.
[8]	彭勃，周文晖，刘济林. 基于Harris角点检测的立体视觉里程计[J]. 兵工学报, 2007, 28(12): 1498-1502.