In the task of the video object tracking, the lack of negative samples and the complex background will result in loss of objects. To solve the above problems, this paper proposes an object tracking algorithm via spatio-temporal context regularized Siamese network. This algorithm adds spatial context information to the Siamese network and uses the information of target object as the positive samples and the information of background as the negative samples to train a tracker model, which is more robust and can suppress the response to the background and highlight the response to the target through regularization constraint. Moreover, the algorithm proposes the mechanism of multi-component matching of the time series, which can dynamically adjust the target template learning rate when the target appearance characteristics are disturbed, thereby ensuring that the target template is not contaminated. Experiments on OTB100 standard dataset show that our algorithm is not only superior to mainstream tracking algorithms in both accuracy and success rate with scores of 0.885 and 0.615 respectively, but also has strong robustness when it faces interference factors such as occlusion, motion blur, illumination changes, background clustering, and fast motion.
其中,mi, ni 分别是由第i帧图像中算法预测的跟踪结果和人工标定值确定的目标矩形框面积。成功率图表示重叠率大于阈值t的帧数占总帧数的百分比。阈值t的取值范围是[0,1],通常取t=0.5。跟踪算法的准确率和成功率均是基于曲线下方面积(area under the curve,AUC)得分计算得到。本实验采取的方式是一次运行评估策略(one-pass evaluation),即所有测试视频只运行一次。
YANGH, SHAOL, ZHENGF, et al. Recent advances and trends in visual tracking: A review [J]. Neurocomputing, 2011, 74(18):3823-3831. DOI:10.1016/j.neucom.2011.07.024 .
HO J, LEE K, YANGM, et al. Visual tracking using learned linear subspaces [C]// IEEE Computer Society Conference on Computer Vision & Pattern Recognition. New York: IEEE Press, 2004:782-789.DOI:10.1109/CVPR.2004.1315111 .
[4]
HUW M, LIX, LUOW H, et al. Single and multiple object tracking using log-euclidean riemannian subspace and block-division appearance model [J]. IEEE Transactions on Pattern Analysis & Machine Intelligence, 2012, 34(12):2420-2440. DOI:10.1109/TPAMI.2012.42 .
[5]
ROSSD A, LIM J, LINR S, et al. Incremental learning for robust visual tracking[J]. International Journal of Computer Vision,2008, 77(1): 125-141. DOI:10.1007/s11263-007-0075-7 .
[6]
MEIX, LINGH. Robust visual tracking using l1 minimization [C] // Proceedings of the 2009 IEEE International Conference on Computer Vision. New York: IEEE Press, 2009: 1436-1443.DOI:10.1109/ICCV.20 .
[7]
GRABNERH, LEISTNERC, BISCHOFH. Semi-supervised on-line boosting for robust tracking [C]// ECCV 08 Proceedings of the 10th European Conference on Computer Vision. Heidelberg:Springer⁃Verlag, 2008:234-247.DOI: 10.1007/978-3-540-88682-2_19 .
[8]
ZHANGZ, WARRELLJ, TORRP H S. Proposal generation for object detection using cascaded ranking SVMs [C]// 2011 IEEE Conference on Computer Vision and Pattern Recognition. Washington D C: IEEE Computer Society, 2011:1497-1504.DOI:10.1109/CVPR.2011.5995411 .
[9]
SUNH, LIJ, CHANGJ, et al. Efficient compressive sensing tracking via mixed classifier decision [J]. Science China Information Sciences, 2016, 59(7):72-102. DOI:10.1007/s11432-015-5424-5 .
[10]
SUNH, LIJ, DUB, et al. On combining compressed sensing and sparse representations for object tracking[C]// PCM 2016 17th Pacific⁃Rim Conference on Advances in Multimedia Information Processing. Heidelberg: Springer⁃Verlag, 2016:32-43.DOI: 10.1007/978-3-319-48890-5_4 .
[11]
HENRIQUESJ F, RUIC, MARTINSP, et al. High-speed tracking with kernelized correlation filters [J]. IEEE Transactions on Pattern Analysis & Machine Intelligence, 2015, 37(3):583-596. DOI: 10.1109/tpami.2014.2345390 .
SUNH, LIJ, DUB, et al. Correlation filtering target tracking based on online multi-lifespan learning [J]. Acta Electronica Sinica, 2017, 45(10):2337-2342. DOI: CNKI:SUN:DZXU.0.2017-10-004(Ch).
[14]
RENS, HEK, GIRSHICKR, et al. Faster R-CNN: Towards real-time object detection with region proposal networks[C]// NIPS 15 Proceedings of the 28th International Conference on Neural Information Processing Systems. Cambridge: MIT Press, 2015:91-99.DOI: 10.1109/TPAMI.2016.2577031 .
[15]
REDMONJ, DIVVALAS, GIRSHICKR, et al. You only look once: Unified, real-time object detection [C]// 2016 IEEE Conference on Computer Vision and Pattern Recognition. New York: IEEE Press, 2016:779-788.DOI:10.1109/CVPR.2016.91 .
[16]
MAC, HUANGJ B, YANGX, et al. Hierarchical convolutional features for visual tracking [C]// ICCV 15 Proceedings of the 2015 IEEE International Conference on Computer Vision. New York: IEEE Press, 2016:3074-3082.DOI:10.1109/ICCV.2015.352 .
[17]
QIY, ZHANGS, QINL, et al. Hedged deep tracking[C]// 2016 IEEE Conference on Computer Vision and Pattern Recognition. New York: IEEE Press, 2016:4303-4311.DOI: 10.1109/CVPR.2016.466 .
[18]
DANELLJANM, ROBINSONA, KHANF S, et al. Beyond correlation filters: Learning continuous convolution operators for visual tracking[C]// European Conference on Computer Vision (LNCS 9909). Heidelberg: Springer⁃Verlag, 2016: 472-488.DOI:10.1007/978-3-319-46454-1_29 .
[19]
DANELLJANM, BHATG, KHANF S, et al. ECO: Efficient convolution operators for tracking [C]// 2017 IEEE Conference on Computer Vision and Pattern Recognition. Washington D C: IEEE Computer Society, 2017:6931-6939.DOI:10.1109/CVPR.2017.733 .
KRISTANM, LEONARDISA, MATASJ, et al. The visual object tracking VOT2017 challenge results[C]// 2017 IEEE International Conference on Computer Vision Workshop. Washington D C: IEEE Computer Society, 2017:1949-1972.DOI:10.1109/ICCVW.2017.230 .
[22]
XUY Z, YUG Z, WANGY P, et al. Car detection from low-altitude UAV Imagery with the faster R-CNN[J]. Journal of Advanced Transportation, 2017, 2017:1-10. DOI:10.1155/2017/2823617 .
[23]
RENS, HEK, GIRSHICKR, et al. Object detection networks on convolutional feature maps [J]. IEEE Transactions on Pattern Analysis & Machine Intelligence, 2017, 39(7):1476-1481. DOI:10.1109/TPAMI.2016.2601099 .
GUANH, XUEX Y, ANZ Y. Advances on application of deep learning for video object tracking [J]. Acta Automatica Sinica, 2016, 42(6):834-847. DOI:10.16383/j.aas.2016.c150705(Ch).
GAOJ Y, YANGX S, ZHANGT Z, et al. Robust visual tracking method via deep learning [J]. Chinese Journal of Computers, 2016, 39(7):1419-1434. DOI:10.11897/SP.J.1016.2016.01419(Ch).
[28]
WANGN, YEUNGD Y. Learning a deep compact image representation for visual tracking[C]// NIPS 13 Proceedings of the 26th International Conference on Neural Information Processing Systems. New York: ACM Press, 2013:809-817.DOI: http://respository.ust.hk/ir/Record/1783.1-61168 .
[29]
WANGL J, OUYANGW L, WANGX G, et al. Visual tracking with fully convolutional networks[C]//2015 IEEE International Conference on Computer Vision. New York: IEEE Press, 2015:3119-3127.DOI: 10.1109/ICCV.2015.357 .
[30]
DANELLJANM, HAGERG, KHANF S, et al. Convolutional features for correlation filter based visual tracking [C]// 2015 IEEE International Conference on Computer Vision Workshop. New York: IEEE Press, 2015:621-629.DOI: 10.1109/ICCVW.2015.84 .
[31]
HOMGS, YOUT, KWAKS, et al. Online tracking by learning discriminative saliency map with convolutional neural network [J]. Computer Science, 2015, 37(5):597-606. DOI: 10.1109/TIP.2015.2510583 .
[32]
RUSSAKOVSKYO, DENGJ, SUH, et al. ImageNet large scale visual recognition challenge [J]. International Journal of Computer Vision, 2015, 115(3):211-252.
[33]
BERTINETTOL, VALMADREJ, HENRIQUESJ F, et al. Fully-convolutional Siamese networks for object tracking [C]// European Conference on Computer Vision(LNCS 9914). Heidelberg: Springer⁃Verlag, 2016:850-865.DOI: 10.1007/978-3-319-48881-3_56 .
[34]
HELDD, THRUNS, SAVARESS. Learning to track at 100 fps with deep regression networks [C]//European Conference on Computer Vision (LNCS 9905). Heidelberg: Springer⁃Verlag, 2016:749-765.DOI: 10.1007/978-3-319-46448-0_45 .
[35]
CHENH, LUCRYS, RAMANAND. Learning policies for adaptive tracking with deep feature cascades[C]// 2017 IEEE International Conference on Computer Vision. Washington D C: IEEE Computer Society, 2017:105-114.DOI:10.1109/ICCV.2017.21 .
[36]
VALMADREJ, BERTINETTOL, HENRIQUESJ, et al. End-to-end representation learning for correlation filter based tracking [C]// 2017 IEEE Conference on Computer Vision and Pattern Recognition. Washington D C: IEEE Computer Society, 2017:5000-5008.DOI:10.1109/CVPR.2017.531 .
[37]
TAOR, GAVVESE, SMEULDERSA W M. Siamese instance search for tracking [C]// 2016 IEEE Conference on Computer Vision and Pattern Recognition. Washington D C: IEEE Computer Society, 2016:1420-1429.DOI: 10.1109/CVPR.2016.158 .
[38]
WANGQ, TENGZ, XINGJ L, et al. Learning attentions: Residual attentional Siamese network for high performance online visual tracking [C]// 2018 IEEE Conference on Computer Vision and Pattern Recognition. Washington D C: IEEE Computer Society, 2018:678-680.DOI: 10.1109/CVPR.2018.00510 .
[39]
HENRIQUESJ F, RUIC, MARTINSP, et al. High-speed tracking with kernelized correlation filters [J]. IEEE Transactions on Pattern Analysis & Machine Intelligence, 2015, 37(3):583-596. DOI: 10.1109/tpami.2014.2345390 .
[40]
MUELLERM, SMITHN, GHANEMB. Context-aware correlation filter tracking [C]// 2017 IEEE Conference on Computer Vision and Pattern Recognition. Washington D C: IEEE Computer Society, 2017:1387-1395.DOI:10.1109/CVPR.2017.152 .
[41]
DANELLJANM, HAGERG, KHANF S, et al. Accurate Scale Estimation for Robust Visual Tracking [EB/OL]. [2018-01-06].
[42]
ZHANGK H, ZHANGL, YANGM H, et al. Fast Tracking via Spatio-Temporal Context Learning [EB/OL]. [2018-03-05].
[43]
BIBIA, MUELLERM, GHANEMB. Target response adaptation for correlation filter tracking[C]// European Conference on Computer Vision (LNCS 9910). Heidelberg: Springer⁃Verlag, 2017:419-433.DOI: 10.1007/978-3-319-46466-4_25 .
[44]
DANELLJANM, HAGERG, KHANF S, et al. Adaptive decontamination of the training set: A unified formulation for discriminative visual tracking [C]// 2016 IEEE Conference on Computer Vision and Pattern Recognition. Washington D C: IEEE Computer Society, 2016:1430-1438.DOI:10.1109/CVPR.2016.159 .
[45]
DANELLJANM, HAGERG, KHANF S, et al. Learning spatially regularized correlation filters for visual tracking [C]// 2018 IEEE Conference on Computer Vision and Pattern Recognition. Washington D C: IEEE Computer Society, 2018:1353-1356.DOI:10.1109/ICCV.2015.490 .
[46]
BERTINETTOL, VALMADREJ, GOLODETZS, et al. Staple: Complementary learners for real-time tracking[C]// 2016 IEEE Conference on Computer Vision and Pattern Recognition. New York: IEEE Press, 2016:1401-1409.DOI: 10.1109/CVPR.2016.156 .
[47]
WANGQ, GAOJ, XINGJ, et al. DCFNet: Discriminant Correlation Filters Network for Visual Tracking [EB/OL].[2017-04-13].
[48]
MAC, YANGX, ZGANGC, et al. Long-term correlation tracking[C]// 2015 IEEE Conference on Computer Vision and Pattern Recognition. Washington D C: IEEE Computer Society, 2015:5388-5396.DOI: 10.1109/CVPR.2015.7299177 .
[49]
ZHANGJ, MAS, SCLAROFFS. MEEM: Robust tracking via multiple experts using entropy minimization[C]// 2014 European Conference on Computer Vision. Heidelberg: Springer⁃Verlag, 2014:188-203.DOI: 10.1007/978-3-319-10599-4_13 .