scholarly journals Online Model Updating and Dynamic Learning Rate-Based Robust Object Tracking

Sensors ◽  
2018 ◽  
Vol 18 (7) ◽  
pp. 2046 ◽  
Author(s):  
Md Islam ◽  
Guoqing Hu ◽  
Qianbo Liu
Keyword(s):  
Object Tracking ◽  
Model Updating ◽  
Learning Rate ◽  
Dynamic Learning

scholarly journals A Fast Object Tracker Based on Integrated Multiple Features and Dynamic Learning Rate

2018 ◽  
Vol 2018 ◽  
pp. 1-14 ◽  
Author(s):  
Jianming Zhang ◽  
You Wu ◽  
Xiaokang Jin ◽  
Feng Li ◽  
Jin Wang
Keyword(s):  
Object Tracking ◽  
Large Scale ◽  
Model Updating ◽  
Color Naming ◽  
Learning Rate ◽  
Appearance Model ◽  
Dynamic Learning ◽  
Tracking Algorithms ◽  
Single Feature ◽  
Confidence Degree

Object tracking is a vital topic in computer vision. Although tracking algorithms have gained great development in recent years, its robustness and accuracy still need to be improved. In this paper, to overcome single feature with poor representation ability in a complex image sequence, we put forward a multifeature integration framework, including the gray features, Histogram of Gradient (HOG), color-naming (CN), and Illumination Invariant Features (IIF), which effectively improve the robustness of object tracking. In addition, we propose a model updating strategy and introduce a skewness to measure the confidence degree of tracking result. Unlike previous tracking algorithms, we judge the relationship of skewness values between two adjacent frames to decide the updating of target appearance model to use a dynamic learning rate. This way makes our tracker further improve the robustness of tracking and effectively prevents the target drifting caused by occlusion and deformation. Extensive experiments on large-scale benchmark containing 50 image sequences show that our tracker is better than most existing excellent trackers in tracking performance and can run at average speed over 43 fps.

Global Convergence of Online BP Training With Dynamic Learning Rate

2012 ◽  
Vol 23 (2) ◽  
pp. 330-341 ◽  
Author(s):  
Rui Zhang ◽  
Zong-Ben Xu ◽  
Guang-Bin Huang ◽  
Dianhui Wang
Keyword(s):  
Global Convergence ◽  
Learning Rate ◽  
Dynamic Learning

scholarly journals A Water Flow Control with OnLine Neurofuzzy Controller using a Dynamic Learning Rate

2018 ◽  
Author(s):  
Hugo Gomes ◽  
Carolina Ferreira1 ◽  
Everthon Oliveira1 ◽  
Agnaldo Reis ◽  
Walmir Caminhas
Keyword(s):  
Flow Control ◽  
Water Flow ◽  
Pid Controller ◽  
Control Process ◽  
Learning Rate ◽  
Dynamic Learning ◽  
The Difference ◽  
Real Plant

This paper describes an ONFC (OnLine Neurofuzzy Controller) application with a dynamic learning rate to control the water flow of a real plant. A revision of ONFC is presented and the ONFCDw version is used, which has an action that minimizes the increase in the difference between the controller weights. The dynamic learning rate used to update the controller weights is described and the results of experiments performed in a water flow control process are presented, comparing the results with the PID controller used in the process.

Fault Diagnosis of Rolling Bearing Based on CNN with Attention Mechanism and Dynamic Learning Rate

2021 ◽  
Author(s):  
Qingyu Zhang ◽  
Hao Wu ◽  
Jinxin Tao ◽  
Wanmeng Ding ◽  
Jinfeng Zhang ◽  
...  
Keyword(s):  
Fault Diagnosis ◽  
Rolling Bearing ◽  
Attention Mechanism ◽  
Learning Rate ◽  
Dynamic Learning

Corrected Background-weighted Histogram Mean Shift Tracking Algorithm Based on Adaptive Bandwidth

2013 ◽  
Vol 401-403 ◽  
pp. 1543-1546
Author(s):  
Feng Liu ◽  
Chao Zhang ◽  
Xiao Pei Wu
Keyword(s):  
Probability Distribution ◽  
Object Tracking ◽  
Model Updating ◽  
Mean Shift ◽  
Spatial Localization ◽  
Target Localization ◽  
Invariant Moment ◽  
Adaptive Kernel ◽  
Weighted Model ◽  
Weighted Histogram

The CBWH (corrected background-weighted histogram) scheme can effectively reduce backgrounds interference in target localization. But it still has the problem of scale and spatial localization inaccuracy. To solve the above issues, we proposed a method which generates a color probability distribution by taking advantage of the targets salient features. In the binary image, we calculate the invariant moment and thus to resize the tracking window of the next frame. A simple background-weighted model updating method is adopted to adapt to the complex background in tracking. Experimental results show that the proposed algorithm improves the robustness of object tracking by self-adaptive kernel-bandwidth updating.

Sign in / Sign up

Export Citation Format

Share Document