引入再检测机制的孪生神经网络目标跟踪

梁浩; 刘克俭; 刘康; 刘岩俊

doi:10.3788/OPE.20192707.1621

您当前的位置：

首页 >

文章列表页 >

引入再检测机制的孪生神经网络目标跟踪

信息科学 | 更新时间：2020-08-13

- 引入再检测机制的孪生神经网络目标跟踪
- Siamese network tracking with redetection mechanism
- 光学精密工程 2019年27卷第7期页码：1621-1631
- 作者机构：
  
  1.中国科学院长春光学精密机械与物理研究所，吉林长春 130033
  2.中国科学院大学，北京 100049
  3.中国人民公安大学，北京 100038
- 作者简介：
  
  [ "梁浩(1994-)，男，河南驻马店人，硕士研究生，2016年于华中科技大学获得学士学位，主要从事计算机视觉及机器学习方面的研究。E-mail:hawkeye.liang@foxmail.com" ]
- 基金信息：
  
  国家重点研发计划资助项目(2016YFC0803000);长春市科技发展计划资助项目(17DY008)
- DOI：10.3788/OPE.20192707.1621
  中图分类号： TP391.4
- 收稿日期：2018-09-21，
  
  录用日期：2018-12-23，
  
  纸质出版日期：2019-07-15
- 稿件说明：
移动端阅览
梁浩, 刘克俭, 刘康, 等. 引入再检测机制的孪生神经网络目标跟踪[J]. 光学精密工程, 2019,27(7):1621-1631.

Hao LIANG, Ke-jian LIU, Kang LIU, et al. Siamese network tracking with redetection mechanism[J]. Optics and precision engineering, 2019, 27(7): 1621-1631.
梁浩, 刘克俭, 刘康, 等. 引入再检测机制的孪生神经网络目标跟踪[J]. 光学精密工程, 2019,27(7):1621-1631. DOI： 10.3788/OPE.20192707.1621.

Hao LIANG, Ke-jian LIU, Kang LIU, et al. Siamese network tracking with redetection mechanism[J]. Optics and precision engineering, 2019, 27(7): 1621-1631. DOI： 10.3788/OPE.20192707.1621.

摘要

针对全卷积孪生神经网络SiamFC在目标快速运动、相似干扰较多等复杂场景下跟踪能力不足的问题，本文引入SINT作为再检测网络对SiamFC进行了改进。本文算法在跟踪响应图出现较多波峰时，启用精确度更高的再检测网络对波峰位置进行重新判定。同时，本文采用了生成式模型构建模板来适应目标的各种变化，以及高置信度的模型更新策略来防止每帧更新可能对模板带来的污染。在OTB2013上对算法性能进行了测试，并选取了9个主流的目标跟踪算法进行对比，本文算法的跟踪精确度达到了88.8%，排名第一，成功率达到了63.2%，排名第二，相比SiamFC有很大地提升。对不同视频序列的分析结果表明，本文算法在目标快速运动、严重遮挡、背景杂波、光照变化和长期跟踪等场景下具有较强的准确性和鲁棒性。

Abstract

To solve the insufficient tracking capability problem for a fully convolutional Siamese network (SiamFC) in complex scenarios such as those involving fast motion and large similar interference

SINT was introduced as a redetection network to improve the SiamFC. When multiple peaks appeared in the tracking response map

the proposed algorithm enabled the redetection network to redetermine the target position with higher accuracy. At the same time

a generative model was adopted to construct a template to adapt to various appearance changes of the target

and a high-confidence model update strategy was used to avoid the model corruption problem. Our algorithm is tested on OTB2013

and nine state-of-the-art algorithms are selected for comparison. The tracking accuracy of our algorithm reaches 88.8%

the best among all the algorithms selectes for comparison

and the success rate reaches 63.2%

which is the second best. Both these properties offer considerable improvement over the SiamFC results. Analysis of several representative video sequences demonstrate that our algorithm has high accuracy and strong robustness in cases involving fast motion

severe occlusion

background clutter

illumination changes

and long-term tracking.

关键词

Keywords

references

程帅, 孙俊喜, 曹永刚, 等.增量深度学习目标跟踪[J].光学精密工程, 2015, 23(4): 1161-1170.

CHENG SH, SUN J X, CAO Y G, et al. . Target tracking based on incremental deep learning[J]. Opt. Precision Eng. , 2015, 23(4): 1161-1170. (in Chinese)

MA C, HUANG J B, YANG X K, et al. . Hierarchical convolutional features for visual tracking[C]//2015 IEEE International Conference on Computer Vision ( ICCV ), December 7-13, 2015. Santiago , Chile . New York , USA : IEEE , 2015: 3074-3082. http://dl.acm.org/citation.cfm?id=2919700

WANG N Y, LI S Y, GUPTA A, et al. . Transferring rich feature hierarchies for robust visual tracking[J]. arXiv preprint arXiv : 1501.04587, 2015 http://www.oalib.com/paper/4069892

DANELLJAN M, ROBINSON A, SHAHBAZ KHAN F, et al. . Beyond Correlation Filters: Learning Continuous Convolution Operators for Visual Tracking [M]//Computer Vision-ECCV 2016. Cham: Springer International Publishing, 2016: 472-488.

NAM H, HAN B. Learning multi-domain convolutional neural networks for visual tracking[C]//2016 IEEE Conference on Computer Vision and Pattern Recognition ( CVPR ), June 27-30, 2016. Las Vegas , NV , USA . New York , USA : IEEE , 2016: 4293-4302. http://www.researchgate.net/publication/283335458_Learning_Multi-Domain_Convolutional_Neural_Networks_for_Visual_Tracking

YUN S, CHOI J, YOO Y, et al. . Action-decision networks for visual tracking with deep reinforcement learning[C]//2017 IEEE Conference on Computer Vision and Pattern Recognition ( CVPR ), July 21-26, 2017. Honolulu , HI . New York , USA : IEEE , 2017. http://www.researchgate.net/publication/319164402_Action-Decision_Networks_for_Visual_Tracking_with_Deep_Reinforcement_Learning

BERTINETTO L, VALMADRE J, HENRIQUES J F, et al. . Fully-convolutional Siamese networks for object tracking [M]//Lecture Notes in Computer Science. Cham: Springer International Publishing, 2016: 850-865.

TAO R, GAVVES E, SMEULDERS A W M. Siamese instance search for tracking[C]//2016 IEEE Conference on Computer Vision and Pattern Recognition ( CVPR ), June 27-30, 2016. Las Vegas , NV , USA . New York , USA : IEEE , 2016: 1420-1429. http://ieeexplore.ieee.org/document/7780527/

DAI K H, WANG Y H, YAN X Y. Long-term object tracking based on Siamese network[C]//2017 IEEE International Conference on Image Processing ( ICIP ), September 17-20, 2017. Beijing . New York , USA : IEEE , 2017: 3640-3644.

KUAI Y L, WEN G J, LI D D. Hyper-feature based tracking with the fully-convolutional Siamese network[C]//2017 International Conference on Digital Image Computing : Techniques and Applications ( DICTA ), November 29- December 1, 2017. Sydney , NSW . New York , USA : IEEE , 2017: 1-7.

KRIZHEVSKY A, SUTSKEVER I, HINTON G E. ImageNet classification with deep convolutional neural networks[J]. Communications of the ACM , 2017, 60(6): 84-90.

GIRSHICK R. Fast r-cnn[C]. Proceedings of the IEEE International Conference on Computer Vision , 2015: 1440-1448.

HARE S, SAFFARI A, TORR P H S. Struck: structured output tracking with kernels[C]//2011 International Conference on Computer Vision , November 6-13, 2011. Barcelona , Spain . New York , USA : IEEE , 2011: 263-270. http://www.researchgate.net/publication/221111405_Struck_Structured_output_tracking_with_kernels?_sg=GWuGKmmglUaEw5YQbw58nUOq5JFhEMoZ4cD7JyUQz5zTEx0302Z-P9nm0xsIbzNPatt9RkbHwgKRA-3xjuddhA

DECLERCQ A, PIATER J H. Online learning of Gaussian mixture models-a two-level approach[C]. VISAPP , 2008: 605-611.

WANG M M, LIU Y, HUANG Z Y. Large margin object tracking with circulant feature maps[C]//2017 IEEE Conference on Computer Vision and Pattern Recognition ( CVPR ), July 21-26, 2017. Honolulu , HI . New York , USA : IEEE , 2017: 21-26. http://www.researchgate.net/publication/315096617_Large_Margin_Object_Tracking_with_Circulant_Feature_Maps

VALMADRE J, BERTINETTO L, HENRIQUES J, et al. . End-to-end representation learning for correlation filter based tracking[C]//2017 IEEE Conference on Computer Vision and Pattern Recognition ( CVPR ), July 21-26, 2017. Honolulu , HI . New York , USA : IEEE , 2017: 5000-5008.

WANG Q, GAO J, XING J L, et al. . Dcfnet: Discriminant correlation filters network for visual tracking[J]. arXiv preprint arXiv : 1704.04057, 2017

MA C, YANG X K, ZHANG C Y, et al. . Long-term correlation tracking[C]//2015 IEEE Conference on Computer Vision and Pattern Recognition ( CVPR ), June 7-12, 2015. Boston , MA , USA . New York , USA : IEEE , 2015: 5388-5396.

BERTINETTO L, VALMADRE J, GOLODETZ S, et al. . Staple: complementary learners for real-time tracking[C]//2016 IEEE Conference on Computer Vision and Pattern Recognition ( CVPR ), June 27-30, 2016. Las Vegas , NV , USA . New York , USA: IEEE , 2016: 1401-1409. http://www.oalib.com/paper/4016386

DANELLJAN M, H?GER G, SHAHBAZ KHAN F, et al. . Accurate scale estimation for robust visual tracking[C]// Proceedings of the British Machine Vision Conference 2014, Nottingham . British Machine Vision Association , 2014: 1-5.

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 , 2015, 37(3): 583-596.

VEDALDI A, LENC K. Matconvnet: Convolutional neural networks for matlab[C]. Proceedings of the 23 rd ACM International Conference on Multimedia , 2015: 689-692. http://www.oalib.com/paper/4067626

JIA Y Q, SHELHAMER E, DONAHUE J, et al. . Caffe: Convolutional architecture for fast feature embedding[C]. Proceedings of the 22 nd ACM International Conference on Multimedia , 2014: 675-678. http://www.oalib.com/paper/4082099

浏览量

123

下载量

CSCD

文章被引用时，请邮件提醒。

提交

工具集

关联资源

基于知识蒸馏的Transformer视觉跟踪器

针对RGBT跟踪的特殊属性的跨模态交互融合网络

基于显著性区域加权的相关滤波目标跟踪

基于双模型的相关滤波跟踪算法

基于双模型的相关滤波跟踪