梯度导向的通道选择目标跟踪算法

doi:10.3778/j.issn.1673-9418.2010029

计算机科学与探索 ›› 2022, Vol. 16 ›› Issue (3): 649-660.DOI: 10.3778/j.issn.1673-9418.2010029

梯度导向的通道选择目标跟踪算法

程世龙, 谢林柏⁺(), 彭力

物联网技术应用教育部工程研究中心（江南大学物联网工程学院）,江苏无锡 214122

收稿日期:2020-10-12 修回日期:2020-11-30 出版日期:2022-03-01 发布日期:2020-12-08
通讯作者: + E-mail: xie_linbo@jiangnan.edu.cn
作者简介:程世龙（1995—）,男,安徽宿州人,硕士研究生,主要研究方向为目标跟踪、深度学习。
谢林柏（1973—）,男,湖南永州人,博士,教授,博士生导师,CAA会员,主要研究方向为过程建模与控制、智能检测与系统安全性。
彭力（1967—）,男,河北唐山人,博士,教授,博士生导师,CAAI会员,CCF会员,主要研究方向为视觉物联网、智能检测。
基金资助:
国家自然科学基金(61873112)

Gradient-Guided Object Tracking Algorithm with Channel Selection

CHENG Shilong, XIE Linbo⁺(), PENG Li

Engineering Research Center of Internet of Things Technology Applications (School of Internet of Things Engineering, Jiangnan University), Ministry of Education, Wuxi, Jiangsu 214122, China

Received:2020-10-12 Revised:2020-11-30 Online:2022-03-01 Published:2020-12-08
About author:CHENG Shilong, born in 1995, M.S. candidate. His research interests include visual object track-ing and deep learning.
XIE Linbo, born in 1973, Ph.D., professor, Ph.D. supervisor, member of CAA. His research interests include process modeling and control, intelligent detection and system safety.
PENG Li, born in 1967, Ph.D., professor, Ph.D. supervisor, member of CAAI and CCF. His research interests include visual Internet of things and intelligent detection.
Supported by:
National Natural Science Foundation of China(61873112)

摘要/Abstract

摘要：

通常在目标跟踪任务中需要跟踪的目标物体具有任意性,同时目标周围可能有相似的干扰物体,这常常导致预训练网络提取的目标特征并不完全适用于当前需要跟踪的目标物体。针对以上问题,在Siamese孪生网络目标跟踪框架下,提出一种新型的基于梯度导向的通道选择目标跟踪算法。首先从预训练网络提取待跟踪目标特征,利用提出的开关-惩罚损失函数对背景中的相似性干扰物体施加惩罚操作,以排除相似物体对跟踪目标的干扰;其次在特征通道选择阶段,根据损失函数反向传播的梯度信息选择特征表达性最强的特征通道;最后在模板分支与搜索分支进行互相关操作部分,利用逐通道互相关方法获得加权的分数响应图以获得更精确的目标位置。在OTB和VOT公开数据集上将该算法和主流算法进行比较。实验结果表明,该算法具有良好的抗背景干扰能力和鲁棒性,在主要跟踪指标上达到了主流跟踪算法的性能。

关键词: 目标跟踪, 梯度导向网络, 惩罚函数, 逐通道互相关

Abstract:

In object tracking task, the target object to be tracked is arbitrary, and there may be similar distractor around the target, which often leads to the target features extracted by the pre-trained network not fully applicable to the tracked target. To solve the above-mentioned issues, the gradient-guided object tracking algorithm is proposed in the Siamese tracking framework. Firstly, the pre-trained network is used to extract the features of object. To eliminate the interference of similar objects, the switch-penalty loss function is used to impose penalty operation on similar objects in the background. Secondly, in the feature channel selection stage, the most expressive feature channels are selected according to the gradient information of back propagation in loss function. Finally, in the part of cross correlation between template branch and search branch, the accurate target position is obtained by using multi-channel cross correlation of the weighted response score map. The proposed algorithm is compared with the mainstream algorithms on OTB and VOT public datasets. Experimental results show that the proposed algorithm has good anti-background interference ability and robustness. The algorithm achieves the performance of the mainstream tracking algorithms in the main tracking indicators.

Key words: object tracking, gradient-guided network, penalty function, multi-channel cross correlation

中图分类号:

TP391.41

程世龙, 谢林柏, 彭力. 梯度导向的通道选择目标跟踪算法[J]. 计算机科学与探索, 2022, 16(3): 649-660.

CHENG Shilong, XIE Linbo, PENG Li. Gradient-Guided Object Tracking Algorithm with Channel Selection[J]. Journal of Frontiers of Computer Science and Technology, 2022, 16(3): 649-660.

图/表 13

图1 目标跟踪算法框架图

Fig.1 Framework of object tracking algorithm

图2 特征热力图

Fig.2 Heat maps of features

图3 逐通道互相关比较图

Fig.3 Comparison of multi-channel cross correlation

图4 搜索分支特征通道选择

Fig.4 Selection of feature channels in search branch

图5 惩罚函数随 α的变化情况

Fig.5 Curve of penalty function with α

表1 OTB数据集上各算法的AUC值

Table 1 AUC of algorithms on OTB dataset

算法	OTB50		OTB100
算法	精度	成功率	精度	成功率
SRDCF^[18]	0.731	0.539	0.789	0.598
DeepSRDCF^[19]	0.772	0.560	0.851	0.635
Staple^[20]	0.683	0.506	0.783	0.578
fDSST^[21]	0.616	0.460	0.686	0.517
CFNet^[22]	0.724	0.535	0.778	0.587
SiamFC^[6]	0.693	0.519	0.772	0.587
SiamRPN^[7]	0.806	0.583	0.847	0.629
DaSiamRPN^[23]	0.829	0.604	0.880	0.658
SiamDWfc^[24]	0.775	0.569	0.828	0.627
GradNet^[25]	0.825	0.597	0.861	0.639
Proposed	0.835	0.619	0.865	0.659

图6 算法在OTB50和OTB100数据集上的评测结果

Fig.6 Evaluation results of algorithms on OTB50 and OTB100 datasets

表2 VOT2018数据集上各跟踪算法性能比较

Table 2 Performance comparison of tracking algorithms on VOT2018 dataset

算法	A	EAO	R
UCT^[27]	0.482	0.206	0.490
UNet-SiamFC^[28]	0.490	0.214	0.580
SiamRPN^[7]	0.490	0.244	0.460
ECO_HC^[29]	0.494	0.177	0.571
SA-Siam^[30]	0.500	0.236	0.459
SimpleRT-MDNet^[31]	0.508	0.218	0.464
Proposed	0.509	0.245	0.460

图7 不同算法在OTB100部分视频序列上可视化跟踪结果

Fig.7 Visualization tracking results on partial video sequences of OTB100 by different algorithms

表3 OTB100上部分视频序列属性

Table 3 Some video properties on OTB100

序列	帧数	主要挑战
Matrix	100	遮挡、光照变化、快速运动、背景混乱
Bird1	408	遮挡、形变、快速运动、超出视野
Box	1 161	遮挡、旋转、尺度变化、运动模糊
MotorRolling	164	旋转、运动模糊、快速运动、背景混乱
Soccer	392	遮挡、运动模糊、快速运动、背景混乱

图8 OTB100上6种不同属性视频序列跟踪成功率

Fig.8 Success rate of video sequences with 6 different attributes on OTB100

表4 OTB100数据集上算法各模块跟踪精度和跟踪成功率对比

Table 4 Tracking accuracy and success rate of each module on OTB100 dataset

模块	精度	成功率
Initial	0.835	0.619
S-P	0.860	0.648
S-P+M-C	0.865	0.659

图9 OTB100上各模块表现情况

Fig.9 Performance of each module on OTB100

参考文献 31

[1]	卢湖川, 李佩霞, 王栋. 目标跟踪算法综述[J]. 模式识别与人工智能, 2018, 31(1): 61-76.
	LU H C, LI P X, WANG D. Visual object tracking: a survey[J]. Pattern Recognition and Artificial Intelligence, 2018, 31(1): 61-76.
[2]	尹宽, 李均利, 李丽, 等. 复杂情况下自适应特征更新目标跟踪算法[J]. 光学学报, 2019, 39(11): 227-242.
	YIN K, LI J L, LI L, et al. Adaptive feature update object-tracking algorithm in complex situations[J]. Acta Optica Sinica, 2019, 39(11): 227-242.
[3]	MA C, HUANG J B, YANG X K, et al. Hierarchical convolu-tional features for visual tracking[C]// Proceedings of the 2015 IEEE International Conference on Computer Vision, Santiago, Dec 11-18, 2015. Washington: IEEE Computer Society, 2015: 3074-3082.
[4]	ZHU Z, WU W, ZOU W, et al. End to end flow correlation tracking with spatial-temporal attention[C]// Proceedings of the 2018 IEEE Conference on Computer Vision and Pattern Recognition, Salt Lake City, Jun 18-22, 2018. Washington:IEEE Computer Society, 2018: 548-557.
[5]	刘芳, 黄光伟, 路丽霞, 等. 自适应模板更新的鲁棒目标跟踪算法[J]. 计算机科学与探索, 2019, 13(1): 83-96.
	LIU F, HUANG G W, LU L X, et al. Robust target tracking algorithm for adaptive template updating[J]. Journal of Fron-tiers of Computer Science and Technology, 2019, 13(1): 83-96.
[6]	BERTINETTO L, VALMADRE J, HENRIQUES J F, et al. Fully-convolutional siamese networks for object tracking[C]// LNCS 9914: Proceedings of the 14th European Conference on Computer Vision, Amsterdam, Oct 8-10 and 15-16, 2016. Cham: Springer, 2016: 850-865.
[7]	LI B, YAN J J, WU W, et al. High performance visual tracking with siamese region proposal network[C]// Proceedings of the 2018 IEEE Conference on Computer Vision and Pattern Recognition, Salt Lake City, Jun 18-22, 2018. Washington:IEEE Computer Society, 2018: 8971-8980.
[8]	REN S Q, HE K M, GIRSHICK R B, et al. Faster R-CNN: towards real-time object detection with region proposal net-works[C]// Proceedings of the Annual Conference on Neural Information Processing Systems, Montreal, Dec 7-12, 2015. Red Hook: Curran Associates, 2015: 91-99.
[9]	仇祝令, 查宇飞, 朱鹏, 等. 基于孪生神经网络在线判别特征的视觉跟踪算法[J]. 光学学报, 2019, 39(9): 253-261.
	QIU Z L, ZHA Y F, ZHU P, et al. Visual tracking algorithm based on online feature discrimination with siamese net-work[J]. Acta Optica Sinica, 2019, 39(9): 253-261.
[10]	PU S, SONG Y B, MA C, et al. Deep attentive tracking via reciprocative learning[C]// Proceedings of the Annual Con-ference on Neural Information Processing Systems, Montréal, Dec 3-8, 2018. Red Hook: Curran Associates, 2018: 1935-1945.
[11]	SELVARAJU R R, COGSWELL M, DAS A, et al. Grad-CAM: visual explanations from deep networks via gradient based localization[C]// Proceedings of the 2017 IEEE Inter-national Conference on Computer Vision, Venice, Oct 22-29, 2017. Washington: IEEE Computer Society, 2017: 618-626.
[12]	ZHOU B L, KHOSLA A, LAPEDRIZA A, et al. Learning deep features for discriminative location[C]// Proceedings of the 2016 IEEE Conference on Computer Vision and Pattern Recognition, Las Vegas, Jun 27-30, 2016. Washington: IEEE Computer Society, 2016: 2921-2929.
[13]	贾永超, 何小卫, 郑忠龙. 融合重检测机制的卷积回归网络目标跟踪算法[J]. 计算机应用, 2019, 39(8): 2247-2251.
	JIA Y C, HE X W, ZHENG Z L. Object tracking algorithm combining re-detection mechanism and convolutional reg-ression network[J]. Journal of Computer Applications, 2019, 39(8): 2247-2251.
[14]	CHEN K, TAO W B. Convolutional regression for visual tracking[J]. IEEE Transactions on Image Processing, 2018, 27(7): 3611-3620. DOI URL
[15]	RUSSAKOVSKY O, DENG J, SU H, et al. ImageNet large scale visual recognition challenge[J]. International Journal of Computer Vision, 2015, 115(3): 211-252. DOI URL
[16]	SIMONYAN K, ZISSERMAN A. Very deep convolutional net-works for large-scale image recognition[J]. arXiv:1409.1556, 2014.
[17]	WU Y, LIM J, YANG M. Object tracking benchmark[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2015, 37(9): 1834-1848. DOI URL
[18]	DANELLJAN M, HÄGER G, KHAN F S, et al. Learning spatially regularized correlation filters for visual tracking[C]// Proceedings of the 2015 IEEE International Conference on Computer Vision Workshops, Santiago, Dec 7-13, 2015. Washington: IEEE Computer Society, 2015: 4310-4318.
[19]	DANELLJAN M, HÄGER G, KHAN F S, et al. Convolu-tional features for correlation filter based visual tracking[C]// Proceedings of the 2015 IEEE International Conference on Computer Vision Workshops, Santiago, Dec 11-18, 2015. Washington: IEEE Computer Society, 2015: 621-629.
[20]	BERTINETTO L, VALMADRE J, GOLODETZ S, et al. Staple: complementary learners for real-time tracking[C]// Proceedings of the 2016 IEEE Conference on Computer Vi-sion and Pattern Recognition, Las Vegas, June 27-30, 2016. Washington: IEEE Computer Society, 2016: 1401-1409.
[21]	DANELLJAN M, HAGER G, KHAN F S, et al. Discrimin-ative scale space tracking[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017, 39(8): 1561-1575. DOI URL
[22]	VALMADRE J, BERTINETTO L, HENRIQUES J F, et al. End to end representation learning for correlation filter based tracking[C]// Proceedings of the 2017 IEEE Conference on Computer Vision and Pattern Recognition, Honolulu, Jul 21-26, 2017. Washington: IEEE Computer Society, 2017: 5000-5008.
[23]	ZHU Z, WANG Q, LI B, et al. Distractor-aware siamese net-works for visual object tracking[C]// LNCS 11213: Proceed-ings of the 15th European Conference on Computer Vision, Munich, Sep 8-14, 2018. Cham: Springer, 2018: 103-119.
[24]	ZHANG Z P, PENG H W. Deeper and wider siamese net-works for real-time visual tracking[C]// Proceedings of the 2019 IEEE Conference on Computer Vision and Pattern Recognition, Long Beach, Jun 16-20, 2019. Piscataway: IEEE, 2019: 4591-4600.
[25]	LI P X, CHEN B Y, OUYANG W L, et al. GradNet: gradient guided network for visual object tracking[C]// Proceedings of the 2019 IEEE/CVF International Conference on Com-puter Vision, Seoul, Oct 27-Nov 2, 2019. Piscataway: IEEE, 2019: 6162-6170.
[26]	KRISTAN M, LEONARDIS A, MATAS J, et al. The sixth visual object tracking VOT2018 challenge results[C]// LNCS 11129: Proceedings of the 15th European Conference on Computer Vision, Munich, Sep 8-14, 2018. Cham: Springer, 2018: 3-53.
[27]	ZHU Z, HUANG G, ZOU W, et al. UCT: learning unified convolutional networks for real-time visual tracking[C]// Pro-ceedings of the 2017 IEEE International Conference on Com-puter Vision, Venice, Oct 22-29, 2017. Washington: IEEE Computer Society, 2017: 1973-1982.
[28]	ZHANG L C, GONZALEZ-GARCIA A, VAN DE WEIJER J, et al. Learning the model update for siamese trackers[C]// Proceedings of the 2019 IEEE/CVF International Conference on Computer Vision, Seoul, Oct 27-Nov 2, 2019. Piscataway: IEEE, 2019: 4010-4018.
[29]	DANELLJAN M, BHAT G, KHAN F S, et al. ECO: efficient convolution operators for tracking[C]// Proceedings of the 2017 IEEE Conference on Computer Vision and Pattern Recognition, Honolulu, Jul 21-26, 2017. Washington: IEEE Computer Society, 2017: 6931-6939.
[30]	HE A F, LUO C, TIAN X M, et al. A twofold siamese net-work for real-time object tracking[C]// Proceedings of the 2018 IEEE Conference on Computer Vision and Pattern Recognition, Salt Lake City, Jun 18-22, 2018. Washington: IEEE Computer Society, 2018: 4834-4843.
[31]	GAO J, HU W M, LU Y, et al. Recursive least squares estimator-aided online learning for visual tracking[C]// Pro-ceedings of the 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition, Seattle, Jun 13-19, 2020. Piscataway: IEEE, 2020: 7386-7395.

梯度导向的通道选择目标跟踪算法

Gradient-Guided Object Tracking Algorithm with Channel Selection

RichHTML

PDF

可视化

摘要/Abstract

引用本文

使用本文

图/表 13

参考文献 31

相关文章 11

编辑推荐

Metrics

[1]	杨政, 邓赵红, 罗晓清, 顾鑫, 王士同. 利用ELM-AE和迁移表征学习构建的目标跟踪系统[J]. 计算机科学与探索, 2022, 16(7): 1633-1648.
[2]	刘艺, 李蒙蒙, 郑奇斌, 秦伟, 任小广. 视频目标跟踪算法综述[J]. 计算机科学与探索, 2022, 16(7): 1504-1515.
[3]	赵运基, 范存良, 张新良. 融合多特征和通道感知的目标跟踪算法[J]. 计算机科学与探索, 2022, 16(6): 1417-1428.
[4]	李运寰, 闻继伟, 彭力. 高帧率的轻量级孪生网络目标跟踪[J]. 计算机科学与探索, 2022, 16(6): 1405-1416.
[5]	杨金龙，汤玉，张光南. 卷积特征多伯努利视频多目标跟踪算法[J]. 计算机科学与探索, 2019, 13(11): 1945-1957.
[6]	刘芳，黄光伟，路丽霞，王洪娟，王鑫. 自适应模板更新的鲁棒目标跟踪算法[J]. 计算机科学与探索, 2019, 13(1): 83-96.
[7]	张晶，王旭，范洪博. 时空上下文相似性的TLD目标跟踪算法[J]. 计算机科学与探索, 2018, 12(7): 1169-1181.
[8]	孟凡琨，巨永锋，文常保 . 随机网格回归Monte Carlo双UAVs最优目标协调跟踪[J]. 计算机科学与探索, 2017, 11(3): 450-458.
[9]	侯跃恩，李伟光. 结构稀疏表示分类目标跟踪算法[J]. 计算机科学与探索, 2016, 10(7): 1035-1043.
[10]	张继，王洪元. 面向视频目标的快速稀疏编码跟踪算法[J]. 计算机科学与探索, 2012, 6(8): 760-768.
[11]	陈龙，尹彪. 自助银行服务中暴力异常事件检测[J]. 计算机科学与探索, 2012, 6(4): 377-384.