Online Model Updating and Dynamic Learning Rate-Based Robust Object Tracking
Robust visual tracking is a significant and challenging issue in computer vision-related research fields and has attracted an immense amount of attention from researchers. Due to various practical applications, many studies have been done that have introduced numerous algorithms. It is considered to...
Main Authors: | , , |
---|---|
Format: | Article |
Language: | English |
Published: |
MDPI AG
2018-06-01
|
Series: | Sensors |
Subjects: | |
Online Access: | http://www.mdpi.com/1424-8220/18/7/2046 |
id |
doaj-58a75b4848174c1f9d5e7bd14f82bea7 |
---|---|
record_format |
Article |
spelling |
doaj-58a75b4848174c1f9d5e7bd14f82bea72020-11-24T21:18:33ZengMDPI AGSensors1424-82202018-06-01187204610.3390/s18072046s18072046Online Model Updating and Dynamic Learning Rate-Based Robust Object TrackingMd Mojahidul Islam0Guoqing Hu1Qianbo Liu2School of Mechanical and Automotive Engineering, South China University of Technology, Guangzhou 510641, ChinaSchool of Mechanical and Automotive Engineering, South China University of Technology, Guangzhou 510641, ChinaSchool of Mechanical and Automotive Engineering, South China University of Technology, Guangzhou 510641, ChinaRobust visual tracking is a significant and challenging issue in computer vision-related research fields and has attracted an immense amount of attention from researchers. Due to various practical applications, many studies have been done that have introduced numerous algorithms. It is considered to be a challenging problem due to the unpredictability of various real-time situations, such as illumination variations, occlusion, fast motion, deformation, and scale variation, even though we only know the initial target position. To address these matters, we used a kernelized-correlation-filter-based translation filter with the integration of multiple features such as histogram of oriented gradients (HOG) and color attributes. These powerful features are useful to differentiate the target from the surrounding background and are effective for motion blur and illumination variations. To minimize the scale variation problem, we designed a correlation-filter-based scale filter. The proposed adaptive model’s updating and dynamic learning rate strategies based on a peak-to-sidelobe ratio effectively reduce model-drifting problems by avoiding noisy appearance changes. The experiment results show that our method provides the best performance compared to other methods, with a distance precision score of 79.9%, overlap success score of 59.0%, and an average running speed of 74 frames per second on the object tracking benchmark (OTB-2015).http://www.mdpi.com/1424-8220/18/7/2046object trackingmachine learningcorrelation filterocclusion detectionscale adaptationonline model updatingdynamic learning rate |
collection |
DOAJ |
language |
English |
format |
Article |
sources |
DOAJ |
author |
Md Mojahidul Islam Guoqing Hu Qianbo Liu |
spellingShingle |
Md Mojahidul Islam Guoqing Hu Qianbo Liu Online Model Updating and Dynamic Learning Rate-Based Robust Object Tracking Sensors object tracking machine learning correlation filter occlusion detection scale adaptation online model updating dynamic learning rate |
author_facet |
Md Mojahidul Islam Guoqing Hu Qianbo Liu |
author_sort |
Md Mojahidul Islam |
title |
Online Model Updating and Dynamic Learning Rate-Based Robust Object Tracking |
title_short |
Online Model Updating and Dynamic Learning Rate-Based Robust Object Tracking |
title_full |
Online Model Updating and Dynamic Learning Rate-Based Robust Object Tracking |
title_fullStr |
Online Model Updating and Dynamic Learning Rate-Based Robust Object Tracking |
title_full_unstemmed |
Online Model Updating and Dynamic Learning Rate-Based Robust Object Tracking |
title_sort |
online model updating and dynamic learning rate-based robust object tracking |
publisher |
MDPI AG |
series |
Sensors |
issn |
1424-8220 |
publishDate |
2018-06-01 |
description |
Robust visual tracking is a significant and challenging issue in computer vision-related research fields and has attracted an immense amount of attention from researchers. Due to various practical applications, many studies have been done that have introduced numerous algorithms. It is considered to be a challenging problem due to the unpredictability of various real-time situations, such as illumination variations, occlusion, fast motion, deformation, and scale variation, even though we only know the initial target position. To address these matters, we used a kernelized-correlation-filter-based translation filter with the integration of multiple features such as histogram of oriented gradients (HOG) and color attributes. These powerful features are useful to differentiate the target from the surrounding background and are effective for motion blur and illumination variations. To minimize the scale variation problem, we designed a correlation-filter-based scale filter. The proposed adaptive model’s updating and dynamic learning rate strategies based on a peak-to-sidelobe ratio effectively reduce model-drifting problems by avoiding noisy appearance changes. The experiment results show that our method provides the best performance compared to other methods, with a distance precision score of 79.9%, overlap success score of 59.0%, and an average running speed of 74 frames per second on the object tracking benchmark (OTB-2015). |
topic |
object tracking machine learning correlation filter occlusion detection scale adaptation online model updating dynamic learning rate |
url |
http://www.mdpi.com/1424-8220/18/7/2046 |
work_keys_str_mv |
AT mdmojahidulislam onlinemodelupdatinganddynamiclearningratebasedrobustobjecttracking AT guoqinghu onlinemodelupdatinganddynamiclearningratebasedrobustobjecttracking AT qianboliu onlinemodelupdatinganddynamiclearningratebasedrobustobjecttracking |
_version_ |
1726008475218083840 |