结合交并比损失的孪生网络目标跟踪算法研究

doi:10.16182/j.issn1004731x.joss.21-0377

系统仿真学报 ›› 2022, Vol. 34 ›› Issue (09): 1956-1967.doi: 10.16182/j.issn1004731x.joss.21-0377

• 仿真建模理论与方法 • 下一篇

结合交并比损失的孪生网络目标跟踪算法研究

周维(), 刘宇翔(), 廖广平, 马鑫

湘潭大学计算机学院　网络空间安全学院，湖南　湘潭　411105

收稿日期:2021-04-28 修回日期:2021-07-08 出版日期:2022-09-18 发布日期:2022-09-23
通讯作者: 刘宇翔 E-mail:zhou_wei@xtu.edu.cn;494172184@qq.com
作者简介:周维（1978-），男，硕士，副教授，硕导，研究方向为计算机视觉、智能系统等。E-mail：zhou_wei@xtu.edu.cn
基金资助:
湖南省科技计划(2016TP1020);衡阳师范学院智能信息处理与应用湖南省重点实验室开放基金(IIPA20K04)

Siamese Object Tracking Algorithm Combined with the Intersection over Union Loss

Wei Zhou(), Yuxiang Liu(), Guangping Liao, Xin Ma

Xiangtan University, School of Computer Science & School of Cyberspace Science, Xiangtan 411105, China

Received:2021-04-28 Revised:2021-07-08 Online:2022-09-18 Published:2022-09-23
Contact: Yuxiang Liu E-mail:zhou_wei@xtu.edu.cn;494172184@qq.com

摘要/Abstract

摘要：

SiamRPN算法采用Ln范数损失训练边界框预测，未考虑预测框与真值框间交并比(inersection over union, IoU)的关系，导致准确性不足。针对该问题，提出一种结合IoU损失的SiamRPN目标跟踪改进算法。设计了IoU-smooth L1范数联合优化模块，对候选正样本进行IoU损失与smooth L₁范数损失的联合优化；依据回归预测结果，用预测框与真值框的IoU作为权重对正样本进行加权分类预测，增加正样本间的区分度，同时确保分类预测与回归预测的关联性。对比实验结果表明：本文所提改进算法能有效提升跟踪性能。

关键词: 机器视觉, 目标跟踪, 孪生网络, 锚点框, 交并比损失

Abstract:

To improve the accuracy of the object bounding box regression prediction of the SiamRPN,solve the problem of low discrimination of positive samples in classification prediction and the lack of correlation between regression prediction and classification prediction, an improved object tracking algorithm of SiamRPN which combined with IoU(intersection over union) loss is proposed. A joint optimization module of IoU-smooth L1 is designed to optimize the IoU loss of the best positive sample and the smooth L1 loss of other positive samples jointly. According to the regression prediction results, the weighted classification prediction is performed on the positive samples with the weight calculated by the IoU of the prediction box and the truth box, so as to increase the discrimination between the positive samples, while ensuring the correlation between the classification prediction and the regression prediction.The results show that the proposed algorithm can effectively improve the tracking performance.

Key words: machine vision, object tracking, siamese network, anchor boxes, inersection over union loss

中图分类号:

TP391.4

周维, 刘宇翔, 廖广平, 马鑫. 结合交并比损失的孪生网络目标跟踪算法研究[J]. 系统仿真学报, 2022, 34(09): 1956-1967.

Wei Zhou, Yuxiang Liu, Guangping Liao, Xin Ma. Siamese Object Tracking Algorithm Combined with the Intersection over Union Loss[J]. Journal of System Simulation, 2022, 34(09): 1956-1967.

图/表 15

图1

图2

表1

图3

图4

图5

图6

图7

图8

图9

图10

图11

表2

图12

图13

参考文献 24

[1]	张荣, 李伟平, 莫同, 等. 深度学习研究综述[J]. 信息与控制, 2018, 47(4): 385-397,410.
	Zhang Rong, Li Weiping, Mo Tong, et al. Review of Deep Learning[J]. Information and Control, 2018, 47(4): 385-397,410.
[2]	侯志强, 戴铂, 胡丹, 等. 基于感知深度神经网络的视觉跟踪[J]. 电子与信息学报, 2016, 38(7): 1616-1623.
	Hou Zhiqiang, Dai Bo, Hu Dan, et al. Robust Visual Tracking via Perceptive Deep Neural Network[J]. Jounal of Electronics & Information Technology, 2016, 38(7): 1616-1623.
[3]	李大湘, 张振. 基于改进U-Net视网膜血管图像分割算法[J]. 光学学报, 2020, 40(10): 58-66.
	Li Daxiang, Zhang Zhen. Improved U-Net Segmentation Algorithm for the Retinal Blood Vessel Images[J]. Acta Optica Sinica, 2020, 40(10): 58-66.
[4]	郭智, 宋萍, 张义, 等. 基于深度卷积神经网络的遥感图像飞机目标检测方法[J]. 电子与信息学报, 2018, 40(11): 2684-2690.
	Guo Zhi, Song Ping, Zhang Yi, et al. Aircraft Detection Method Based on Deep Convolutional Neural Network for Remote Sensing Images[J]. Jounal of Electronics &Information Technology, 2018, 40(11): 2684-2690.
[5]	Tao R, Gavves E, Smeulders A W M. Siamese Instance Search for Tracking[C]// IEEE Conference on Computer Vision and Pattern Recognition. Las Vegas, USA: IEEE, 2016: 1420-1429.
[6]	Bertinetto L, Valmadre J, Henriques J F, et al. Fully-convolutional Siamese Networks for Object Tracking[C]// European Conference on Computer Vision.Amsterdam, Netherlands: Springer, 2016: 850-865.
[7]	Li B, Yan J, Wu W, et al. High Performance Visual Tracking with Siamese Region Proposal Network[C]//IEEE Conference on Computer Vision and Pattern Recognition. Salt Lake City, USA: IEEE, 2018: 8971-8980.
[8]	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 (S0162-8828), 2016, 39(6): 1137-1149.
[9]	Guo Q, Feng W, Zhou C, et al. Learning Dynamic Siamese Network for Visual Object Tracking[C]// IEEE International Conference on Computer Vision. Venice, Italy: IEEE, 2017: 1763-1771.
[10]	Zhu Z, Wang Q, Li B, et al. Distractor-Aware Siamese Networks for Visual Object Tracking[C]// European Conference on Computer Vision(ECCV). Munich, Germany: Springer, 2018: 101-117.
[11]	Yu J, Jiang Y, Wang Z, et al. Unitbox: An Advanced Object Detection Network[C]// 24th ACM International Conference on Multimedia. Amsterdam, Netherlands:ArXiv e-prints, 2016: 516-520.
[12]	Rezatofighi H, Tsoi N, Gwak J Y, et al. General-lzed Intersection Over Union: A Metric and a Loss for Bounding Box Regression[C]// IEEE Conference on Computer Vision and Pattern Recognition. Long Beach, USA: IEEE, 2019: 658-666.
[13]	Wu Y, Lim J, Yang M H. Online Object Tracking: A Benchmark[C]// IEEE Conference on Computer Vision and Pattern Recognition. Portland, USA: IEEE, 2013: 2411-2418.
[14]	Wu Y, Lim J, Yang M H. Object Tracking Benchmark[J].IEEE Transactions on Pattern Analysis and Machine Intelligence (S0162-8828), 2015, 37(9): 1834-1848.
[15]	Krizhevsky A, Sutskever I, Hinton G E. Imagenet Classification with Deep Convolutional Neural Networks[J]. Communications of the ACM (S0001-0782), 2017, 60(6): 84-90.
[16]	Tychsen-Smith L, Petersson L. Improving Object Localization with Fitness Nms and Bounded Iou Loss[C]// IEEE Conference on Computer Vision and Pattern Recognition. Salt Lake City, USA: IEEE, 2018: 6877-6885.
[17]	Li B, Wu W, Wang Q, et al. Siamrpn++: Evolution of Siamese Visual Tracking with Very Deep Net-Works[C]// IEEE Conference on Computer Vision and Pattern Recognition. Long Beach, USA: IEEE, 2019: 4282-4291.
[18]	Zheng Z, Wang P, Liu W, et al. Distance-IoU Loss: Faster and Better Learning for Bounding Box Regression[C]//AAAI Conference on Artificial Intelligence. New York, USA: ArXiv e-prints, 2020: 12993-13000.
[19]	Huang L, Zhao X, Huang K. Got-10k: A Large High-diversity Benchmark for Generic Object Tracking in the Wild[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence (S0162-8828), 2019, 43(5): 1562-1577.
[20]	Real E, Shlens J, Mazzocchi S, et al. Youtube-Boundingboxes: A Large High-precision Human-Annotated Data Set for Object Detection in Video[C]// IEEE Conference on Computer Vision and Pattern Recognition. Honolulu, USA: IEEE, 2017: 5296-5305.
[21]	Bertinetto L, Valmadre J, Golodetz S, et al. Staple: Complementary Learners for Real-time Tracking[C]//IEEE Conference on Computer Vision and Pattern Recognition. Las Vegas, USA: IEEE, 2016: 1401-1409.
[22]	Henriques J F, Caseiro R, Martins P, et al. High-speed Tracking with Kernelized Correlation Filters[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence (S0162-8828), 2014, 37(3): 583-596.
[23]	Lukezic A, Vojir T, Zajc L C, et al. Discriminative Correlation Filter with Channel and Spatial Reliability[C]// IEEE Conference on Computer Vision and Pattern Recognition. Honolulu, USA: IEEE, 2017: 6309-6318.
[24]	Li F, Tian C, Zuo W, et al. Learning Spatial-Temporal Regularized Correlation Filters for Visual Tracking[C]//IEEE Conference on Computer Vision and Pattern Recognition. Salt Lake City, USA: IEEE, 2018: 4904-4913.

网络层类型	尺寸/步长	卷积核数量
Conv1	11×11/2	96
Pool1	3×3/2
Conv2	5×5/1	256
Pool2	3×3/2
Conv3	3×3/1	384
Conv4	3×3/1	384
Conv5	3×3/1	256

跟踪算法	准确率	鲁棒性	EAO	FPS
Staple	0.531	0.632	0.186	77
KCF	0.467	1.302	0.097	172
CSRDCF	0.505	0.487	0.211	13
STRCF	0.480	0.613	0.162	32
SiamRPN	0.532	0.459	0.247	160
SiamFC	0.492	0.609	0.199	101
SiamIoU	0.521	0.379	0.276	153

结合交并比损失的孪生网络目标跟踪算法研究

Siamese Object Tracking Algorithm Combined with the Intersection over Union Loss

RichHTML

PDF (PC)

可视化

摘要/Abstract

引用本文

使用本文

图/表 15

参考文献 24

相关文章 15

编辑推荐

Metrics

本文评价

[1]	张思贤, 杨艺, 张猛, 米鹏博. 高效的多特征自适应相关滤波跟踪器[J]. 系统仿真学报, 2022, 34(8): 1864-1873.
[2]	袁彬淦, 钟铭恩, 倪晶鑫. 线束端子压接后外观缺陷的视觉检测算法研究[J]. 系统仿真学报, 2022, 34(5): 1152-1159.
[3]	吴定会, 张桐瑞, 张秀丽. 扰动累积下基于数字孪生的车间重调度[J]. 系统仿真学报, 2022, 34(3): 573-583.
[4]	李琪, 墨瀚林, 王向东, 李华. 短道速滑场景下的多目标跟踪与运动仿真[J]. 系统仿真学报, 2021, 33(5): 1039-1050.
[5]	冯晓, 张辉, 周蕊, 乔璐, 魏东, 李丹丹, 张玉尧, 郑国清. 基于深度学习和籽粒双面特征的玉米品种识别[J]. 系统仿真学报, 2021, 33(12): 2983-2991.
[6]	柳有权, 裴雪, 李婉, 刘正雄. 基于邻近目标置信度评估的视觉目标跟踪与定位[J]. 系统仿真学报, 2020, 32(7): 1294-1300.
[7]	张良, 郝凯锋. 基于RSSI变化序列的MAC地址与目标匹配方法[J]. 系统仿真学报, 2020, 32(1): 113-121.
[8]	吴昀璞, 金炜东, 任俊箫. 基于孪生卷积网络的高速列车转向架故障辨识[J]. 系统仿真学报, 2019, 31(11): 2562-2568.
[9]	张青, 邹湘军, 林桂潮, 孙艳辉. 草莓重量和形状图像特征提取与在线分级方法[J]. 系统仿真学报, 2019, 31(1): 7-9.
[10]	孙永侃, 张萍萍. 基于横向地球坐标系的极区超视距目标跟踪[J]. 系统仿真学报, 2018, 30(9): 3502-3513.
[11]	李菊, 曹明伟, 余烨, 夏瑜, 周立凡. 抗遮挡自适应的粒子滤波算法[J]. 系统仿真学报, 2018, 30(9): 3552-3557.
[12]	姜明新, 潘志庚, 王兰芳, 胡铸鑫. 基于深度去噪自编码器的RGB-D视频目标跟踪[J]. 系统仿真学报, 2018, 30(11): 4276-4283.
[13]	张志利, 王超, 龙勇, 王韶迪. 基于核相关滤波器无标识AR的目标跟踪仿真[J]. 系统仿真学报, 2018, 30(11): 4462-4468.
[14]	牛为华, 梁贵书, 赵鹏. 基于稀疏表示和M-ELM的断路器故障诊断模型[J]. 系统仿真学报, 2017, 29(11): 2828-2839.
[15]	陈金广, 江梦茜, 马丽丽. 存在目标交叉情形的扩展目标跟踪算法[J]. 系统仿真学报, 2017, 29(1): 7-13.