Research on single object tracking algorithm based on Siamese network and Kalman filter

Yangyang Zhang; Chong Guo; Hongbo Zhu

doi:10.1117/12.2675178

25 May 2023 Research on single object tracking algorithm based on Siamese network and Kalman filter

Yangyang Zhang, Chong Guo, Hongbo Zhu

Proceedings Volume 12636, Third International Conference on Machine Learning and Computer Application (ICMLCA 2022); 126364Y (2023) https://doi.org/10.1117/12.2675178
Event: Third International Conference on Machine Learning and Computer Application (ICMLCA 2022), 2022, Shenyang, China

Abstract

The single object tracking algorithm based on Siamese Network transforms the object tracking problem into the similarity calculation problem of features. By inputting the template and the search frame into the Siamese Network with the same structure and sharing weights, the corresponding features are extracted respectively, and the feature map is output. Then, convolution operation is performed on the feature map to find the area where the search frame is most similar to the template frame, so as to achieve the tracking of a single object. However, this method only uses the appearance feature of the object to track the object, and does not use the motion feature of the object to predict the position of the object. In a scene with multiple objects with similar appearance, it is easy to overlap the spatial positions of similar objects. Or the tracking frame drift phenomenon occurs due to occlusion or other reasons, resulting in the loss of tracking of the object. Kalman Filter is a linear system that can predict the next state of the object based on the current state and known information, and continuously predict the object state through continuous iteration. This feature can be used to predict the motion trajectory of the object, and combining it with the Siamese Network can greatly improve the anti-interference ability of the Siamese Network. This paper uses SiamRPN as the basic network and introduces the Kalman Filter algorithm to predict the object trajectory. The purpose of this paper is to fuse the appearance and motion features of the object, and conduct a comparative experiment with the tracking of a single feature to study the impact of different features on the object tracking performance.

Citation Download Citation

Yangyang Zhang, Chong Guo, and Hongbo Zhu "Research on single object tracking algorithm based on Siamese network and Kalman filter", Proc. SPIE 12636, Third International Conference on Machine Learning and Computer Application (ICMLCA 2022), 126364Y (25 May 2023); https://doi.org/10.1117/12.2675178

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available

Members: $17.00

Non-members: $21.00 ADD TO CART

PROCEEDINGS
6 PAGES

DOWNLOAD PAPER SAVE TO MY LIBRARY

GET CITATION

RIGHTS & PERMISSIONS

Get copyright permission Get copyright permission on Copyright Marketplace

KEYWORDS

Signal filtering

Detection and tracking algorithms

Electronic filtering

Tunable filters

Video

Feature extraction

Image processing

Show All Keywords

Keywords/Phrases

Search In:

Publication Years