基于卷积神经网络多层特征融合的目标跟踪Target tracking based on multi-layer feature fusion of convolutional neural network
苗军,李凯,许少武
摘要(Abstract)：
为了对目标对象进行鲁棒的特征表达以用于更加准确的跟踪和定位,针对卷积神经网络的不同层能够提取到目标的不同特征表达这一特性,提出一种基于卷积神经网络多层特征融合的目标跟踪方法。该方法将网络提取到的低层纹理特征和高层语义特征进行有效的融合,并将之用于跟踪网络模型的训练。此外,融合后的特征表达还被用来训练Bounding Box回归模型,用于对跟踪结果的优化。通过在OTB100标准数据集上将所提方法与目前有代表性的几种跟踪方法进行对比,所提出的特征融合方法使系统的综合指标得到了显著提升,证明所提方法的有效性。
关键词(KeyWords)： 目标跟踪;特征融合;特征表达;目标定位;卷积神经网络;回归模型
基金项目(Foundation): 国家自然科学基金项目(61650201);; 北京市自然科学基金项目(4162058);; 北京未来芯片技术高精尖创新中心科研基金(KYJJ2018004);; 北京信息科技大学2018年人才培养质量提高经费(5111823402)~~
作者(Author): 苗军,李凯,许少武
DOI: 10.16652/j.issn.1004-373x.2018.24.028
参考文献(References)：
- [1] WANG N,YEUNG D Y. Learning a deep compact image representation for visual tracking[C]//Proceedings of the 26th International Conference on Neural Information Processing Systems.Lake Tahoe:Curran Associations Inc.,2013:809-817.
- [2] WANG N,LI S,GUPTA A,et al. Transferring rich feature hierarchies for robust visual tracking[J/OL].[2015-04-23]. http://www.docin.com/p-2043914692.html.
- [3] WANG L,OUYANG W,WANG X,et al. Visual tracking with fully convolutional networks[C]//Proceedings of IEEE International Conference on Computer Vision. Santiago:IEEE,2015:3119-3127.
- [4] MA C,HUANG J,YANG X,et al. Hierarchical convolutional features for visual tracking[C]//Proceedings of IEEE International Conference on Computer Vision. Santiago:IEEE,2015:3074-3082.
- [5] HONG S,YOU T,KWAK S,et al. Online tracking by learning discriminative saliency map with convolutional neural network[C]//Proceedings of the 32nd International Conference on Machine Learning. Lille:IEEE,2015:597-606.
- [6] NAM H,HAN B. Learning multi-domain convolutional neural networks for visual tracking[J/OL].[2015-10-27]. https://arxiv.org/pdf/1510.07945v1.pdf.
- [7] WU Y,LIM J,YANG M H. Object tracking benchmark[J].IEEE transactions on pattern analysis&machine intelligence,2015,37(9):1834-1848.
- [8] DANELLJAN M,ROBINSON A,KHAN F S,et al. Beyond correlation filters:learning continuous convolution operators for visual tracking[J/OL].[2016-08-29]. https://arxiv. org/pdf/1608.03773.pdf.
- [9] DANELLJAN M,H?GER G,KHAN F S,et al. Learning spatially regularized correlation filters for visual tracking[C]//Proceedings of IEEE International Conference on Computer Vision. Santiago:IEEE,2015:4310-4318.
- [10] MA C,YANG X,ZHANG C,et al. Long-term correlation tracking[C]//IEEE Conference on Computer Vision and Pattern Recognition.[S.l.]:IEEE,2015:5388-5396.
- [11] HENRIQUES J F,RUI C,MARTINS P,et al. High-speed tracking with kernelized correlation filters[J]. IEEE transactions on pattern analysis&machine intelligence,2015,37(3):583-596.