Research on Single Object Tracking with Attention Mechanism
OUYANG Juntao
GUO Jianhui
Abstract：Object tracking plays an important role in the fields of intelligent security,traffic and monitoring.To make full use of the feature information of data and improve the tracking effect of the tracker,this paper applies the attention mechanism to single object tracking.Specifically,after the feature extraction sub-network extracts the input data features,the attention module is ap-plied to the results of the cross-correlation between template features and search features.The attention module of this paper in-cludes two parts,which are channel attention and spatial attention.The channel attention sub-module performs global average pool-ing on the input data in the width and height dimensions,and represents each channel with a number,and then passes through the fully connected layer.Model each channel and obtain the weight of each channel.The spatial attention sub-module first performs global average pooling on the channel dimension,represents the channel at each position of the feature map with a number,and then obtains the weight of each channel through the self-attention mechanism.Finally the result is applied to the input feature map,and then subsequent operations are performed.The experiment on the GOT10K,OTB100 and LaSOT test sets show that the method can improve the target tracking effectively.
Keywords：single object trackingsiamese networkattention mechanismdeep learning
Publication Date：2025-09-20
Online Publishing Date：2025-12-23(First online date of this platform, not the publication date of the document)
Pages：8( 2465-2471,2508 )