The article points that object tracking is an important research topic in the field of machine vision , and it is widely used in national defense , transportation and other fields .With the development of training data and hard-ware, more and more researchers are applying the technique of deep learning to visual tracking .Recently, a large number of tracking algorithms based on deep learning are proposed .Compared with the traditional machine learning methods, the techniques using convolution neural networks with multiple hidden layers have more powerful capaci -ties of feature learning and feature expression .Then, it analyzes the difficult problems in object tracking and the possibility of using convolution neural networks to solve object-tracking problems .Furthermore , the development of convolution neural networks in visual tracking is reviewed , and the latest results of applying convolution neural net-works to visual target tracking are summarized and analyzed .Finally, the future development of convolutional neu-ral networks in object tracking is discussed .%目标跟踪是机器视觉领域一个重要的研究方向,在军事、交通等领域有着广泛的应用.随着训练数据和硬件的发展,越来越多的学者将深度学习应用于视觉跟踪领域.近几年来,一大批基于深度学习的跟踪算法被提出,与传统的机器学习方法相比,包含多个隐含层的卷积神经网络(CNN)具有更强大的特征学习和特征表达能力.分析了目标跟踪中的难题以及用卷积神经网络解决此类问题的可能性,综述了卷积神经网络在视觉跟踪领域的发展,并对卷积神经网络在视觉目标跟踪中的最新成果进行了总结和深入分析,最后对卷积神经网络在目标跟踪领域未来的发展进行了展望.
展开▼
