Deep Correlation Tracking with Backtracking
Summary :
Feature extractor is an important component of a tracker and the convolutional neural networks (CNNs) have demonstrated excellent performance in visual tracking. However, the CNN features cannot perform well under conditions of low illumination. To address this issue, we propose a novel deep correlation tracker with backtracking, which consists of target translation, backtracking and scale estimation. We employ four correlation filters, one with a histogram of oriented gradient (HOG) descriptor and the other three with the CNN features to estimate the translation. In particular, we propose a backtracking algorithm to reconfirm the translation location. Comprehensive experiments are performed on a large-scale challenging benchmark dataset. And the results show that the proposed algorithm outperforms state-of-the-art methods in accuracy and robustness.
- Publication
- IEICE TRANSACTIONS on Fundamentals Vol.E100-A No.7 pp.1601-1605
- Publication Date
- 2017/07/01
- Publicized
- Online ISSN
- 1745-1337
- DOI
- 10.1587/transfun.E100.A.1601
- Type of Manuscript
- LETTER
- Category
- Vision
Authors
Yulong XU
PLA University of Science and Technology
Yang LI
PLA University of Science and Technology
Jiabao WANG
PLA University of Science and Technology
Zhuang MIAO
PLA University of Science and Technology
Hang LI
PLA University of Science and Technology
Yafei ZHANG
PLA University of Science and Technology
Gang TAO
Anhui Keli Information Industry Co. Ltd.
Keyword
Latest Issue
Copyrights notice of machine-translated contents
The copyright of the original papers published on this site belongs to IEICE. Unauthorized use of the original or translated papers is prohibited. See IEICE Provisions on Copyright for details.
Cite this
Copy
Yulong XU, Yang LI, Jiabao WANG, Zhuang MIAO, Hang LI, Yafei ZHANG, Gang TAO, "Deep Correlation Tracking with Backtracking" in IEICE TRANSACTIONS on Fundamentals,
vol. E100-A, no. 7, pp. 1601-1605, July 2017, doi: 10.1587/transfun.E100.A.1601.
Abstract: Feature extractor is an important component of a tracker and the convolutional neural networks (CNNs) have demonstrated excellent performance in visual tracking. However, the CNN features cannot perform well under conditions of low illumination. To address this issue, we propose a novel deep correlation tracker with backtracking, which consists of target translation, backtracking and scale estimation. We employ four correlation filters, one with a histogram of oriented gradient (HOG) descriptor and the other three with the CNN features to estimate the translation. In particular, we propose a backtracking algorithm to reconfirm the translation location. Comprehensive experiments are performed on a large-scale challenging benchmark dataset. And the results show that the proposed algorithm outperforms state-of-the-art methods in accuracy and robustness.
URL: https://global.ieice.org/en_transactions/fundamentals/10.1587/transfun.E100.A.1601/_p
Copy
@ARTICLE{e100-a_7_1601,
author={Yulong XU, Yang LI, Jiabao WANG, Zhuang MIAO, Hang LI, Yafei ZHANG, Gang TAO, },
journal={IEICE TRANSACTIONS on Fundamentals},
title={Deep Correlation Tracking with Backtracking},
year={2017},
volume={E100-A},
number={7},
pages={1601-1605},
abstract={Feature extractor is an important component of a tracker and the convolutional neural networks (CNNs) have demonstrated excellent performance in visual tracking. However, the CNN features cannot perform well under conditions of low illumination. To address this issue, we propose a novel deep correlation tracker with backtracking, which consists of target translation, backtracking and scale estimation. We employ four correlation filters, one with a histogram of oriented gradient (HOG) descriptor and the other three with the CNN features to estimate the translation. In particular, we propose a backtracking algorithm to reconfirm the translation location. Comprehensive experiments are performed on a large-scale challenging benchmark dataset. And the results show that the proposed algorithm outperforms state-of-the-art methods in accuracy and robustness.},
keywords={},
doi={10.1587/transfun.E100.A.1601},
ISSN={1745-1337},
month={July},}
Copy
TY - JOUR
TI - Deep Correlation Tracking with Backtracking
T2 - IEICE TRANSACTIONS on Fundamentals
SP - 1601
EP - 1605
AU - Yulong XU
AU - Yang LI
AU - Jiabao WANG
AU - Zhuang MIAO
AU - Hang LI
AU - Yafei ZHANG
AU - Gang TAO
PY - 2017
DO - 10.1587/transfun.E100.A.1601
JO - IEICE TRANSACTIONS on Fundamentals
SN - 1745-1337
VL - E100-A
IS - 7
JA - IEICE TRANSACTIONS on Fundamentals
Y1 - July 2017
AB - Feature extractor is an important component of a tracker and the convolutional neural networks (CNNs) have demonstrated excellent performance in visual tracking. However, the CNN features cannot perform well under conditions of low illumination. To address this issue, we propose a novel deep correlation tracker with backtracking, which consists of target translation, backtracking and scale estimation. We employ four correlation filters, one with a histogram of oriented gradient (HOG) descriptor and the other three with the CNN features to estimate the translation. In particular, we propose a backtracking algorithm to reconfirm the translation location. Comprehensive experiments are performed on a large-scale challenging benchmark dataset. And the results show that the proposed algorithm outperforms state-of-the-art methods in accuracy and robustness.
ER -
English
