Combining Siamese Network and Regression Network for Visual Tracking

Yao GE  Rui CHEN  Ying TONG  Xuehong CAO  Ruiyu LIANG  

IEICE TRANSACTIONS on Information and Systems   Vol.E103-D   No.8   pp.1924-1927
Publication Date: 2020/08/01
Publicized: 2020/05/13
Online ISSN: 1745-1361
DOI: 10.1587/transinf.2020EDL8032
Type of Manuscript: LETTER
Category: Image Recognition, Computer Vision
regression network,  siamese network,  two-stage,  visual tracking,  

Full Text: PDF(1.4MB)>>
Buy this Article

We combine the siamese network and the recurrent regression network, proposing a two-stage tracking framework termed as SiamReg. Our method solves the problem that the classic siamese network can not judge the target size precisely and simplifies the procedures of regression in the training and testing process. We perform experiments on three challenging tracking datasets: VOT2016, OTB100, and VOT2018. The results indicate that, after offline trained, SiamReg can obtain a higher expected average overlap measure.