Multi-layer features template update object tracking algorithm based on SiamFC++
Abstract SiamFC++ only extracts the object feature of the first frame as a tracking template, and only uses the highest level feature maps in both the classification branch and the regression branch, so that the respective characteristics of the two branches are not fully utilized. In view of this,...
Main Authors: | Xiaofeng Lu, Xuan Wang, Zhengyang Wang, Xinhong Hei |
---|---|
Format: | Article |
Language: | English |
Published: |
SpringerOpen
2024-01-01
|
Series: | EURASIP Journal on Image and Video Processing |
Subjects: | |
Online Access: | https://doi.org/10.1186/s13640-023-00616-x |
Similar Items
-
Multi‐template temporal information fusion for Siamese object tracking
by: Xiaofeng Lu, et al.
Published: (2023-02-01) -
AtptTrack: Asymmetric Transformer Tracker With Prior Templates
by: Hao Zhang, et al.
Published: (2024-01-01) -
Learning the model update with local trusted templates for visual tracking
by: Zhiyong An, et al.
Published: (2023-02-01) -
AR Long-Term Tracking Combining Multi-Attention and Template Updating
by: Mengru Guo, et al.
Published: (2023-04-01) -
An Anchor-Free Siamese Network with Multi-Template Update for Object Tracking
by: Tongtong Yuan, et al.
Published: (2021-04-01)