摘要
ECO算法虽然在跟踪效果上有了很大的提升,但是它在复杂场景中表现不好甚至会丢失目标,即鲁棒性不高;对此,文章探讨了怎样在ECO算法中利用深度特征处理深层语义的能力和浅层特征处理纹理颜色信息的能力来提升算法的鲁棒性,同时对比了深度特征和浅层特征在目标跟踪的不同作用,并因此提出了改进方法,首先在深度网络上选择了具有更深层次的ResNet-101网络;其次修改了适宜此网络的参数;算法在OTB-2015进行的实验也取得了比较良好的结果,在低分辨率、背景杂波、光照变化及尺度变化4个挑战因素的成功率分别领先基准算法ECO为0.135,0.034,0.031,0.024。
Although ECO algorithm has greatly improved the tracking effect,there are some shortcomings in complex scenes and even loses the target,and its robustness is not high.In this regard,how to improve the robustness of ECO algorithm is discussed by using the ability of deep feature to process deep semantics and the ability of shallow feature to process texture color information.At the same time,the different functions of depth feature and shallow feature are compared in target tracking,and an improved method is put forwarded,Firstly,the ResNet-101 network with deeper level is selected on the deep network;Secondly,the parameters suitable for this network are modified.The experiment of the algorithm in OTB-2015 also achieved good results,The success rates of low resolution,background clutter,illumination change and scale change are 0.135,0.034,0.031 and 0.024 respectively.
作者
王奇
王录涛
江山
文成江
WANG Qi;WANG Lutao;JIANG Shan;WEN Chengjiang(College of Computer Sciences,Chengdu University of Information Technology,Chengdu 610225,China)
出处
《计算机测量与控制》
2022年第6期210-215,共6页
Computer Measurement &Control
基金
四川省科技厅重点研发项目(2020YFG0442)。
关键词
目标跟踪
深度特征
浅层特征
鲁棒性
复杂场景
visual tracking
depth features
shallow features
robustness
complex scene
