Learning Deconvolutional Network for Object Tracking
Object tracking can be tackled by learning a model of tracking the target's appearance sequentially. Therefore, robust appearance representation is a critical step in visual tracking. Recently, deep convolution network has demonstrated remarkable ability in visual tracking via leveraging robust...
Main Authors: | , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
IEEE
2018-01-01
|
Series: | IEEE Access |
Subjects: | |
Online Access: | https://ieeexplore.ieee.org/document/8326476/ |
id |
doaj-ecd0e6b6ea364ec88eb0ac7f6a93b6fb |
---|---|
record_format |
Article |
spelling |
doaj-ecd0e6b6ea364ec88eb0ac7f6a93b6fb2021-03-29T21:00:08ZengIEEEIEEE Access2169-35362018-01-016180321804110.1109/ACCESS.2018.28200048326476Learning Deconvolutional Network for Object TrackingXiankai Lu0https://orcid.org/0000-0002-9543-6960Hong Huo1Tao Fang2Huanlong Zhang3Department of Automation, Shanghai Jiao Tong University, Shanghai, ChinaDepartment of Automation, Shanghai Jiao Tong University, Shanghai, ChinaDepartment of Automation, Shanghai Jiao Tong University, Shanghai, ChinaZhengzhou University of Light Industry, Zhangzhou, ChinaObject tracking can be tackled by learning a model of tracking the target's appearance sequentially. Therefore, robust appearance representation is a critical step in visual tracking. Recently, deep convolution network has demonstrated remarkable ability in visual tracking via leveraging robust high-level features. To obtain these high-level features, convolution and pooling operations are executed alternatively in deep convolution network. However, these operations lead to low spatial resolution feature maps which degrade the localization precision in tracking. While low level features have sufficient spatial resolution, their representation ability is insufficient. To mitigate this issue, we exploited deconvolution network in visual tracking. This deconvolution network works as a learnable upsampling layer which takes low-resolution high-level feature maps as input and outputs enlarged feature maps. Meanwhile, the low level feature maps are fused with these high level feature maps via a summarization operation to better represent target appearance. We formulate the network training as a regression issue and train this network end to end. Extensive experiments on two tracking benchmarks demonstrate the effectiveness of our method.https://ieeexplore.ieee.org/document/8326476/Object trackingdeep learningdeconvolution neural networkregression network |
collection |
DOAJ |
language |
English |
format |
Article |
sources |
DOAJ |
author |
Xiankai Lu Hong Huo Tao Fang Huanlong Zhang |
spellingShingle |
Xiankai Lu Hong Huo Tao Fang Huanlong Zhang Learning Deconvolutional Network for Object Tracking IEEE Access Object tracking deep learning deconvolution neural network regression network |
author_facet |
Xiankai Lu Hong Huo Tao Fang Huanlong Zhang |
author_sort |
Xiankai Lu |
title |
Learning Deconvolutional Network for Object Tracking |
title_short |
Learning Deconvolutional Network for Object Tracking |
title_full |
Learning Deconvolutional Network for Object Tracking |
title_fullStr |
Learning Deconvolutional Network for Object Tracking |
title_full_unstemmed |
Learning Deconvolutional Network for Object Tracking |
title_sort |
learning deconvolutional network for object tracking |
publisher |
IEEE |
series |
IEEE Access |
issn |
2169-3536 |
publishDate |
2018-01-01 |
description |
Object tracking can be tackled by learning a model of tracking the target's appearance sequentially. Therefore, robust appearance representation is a critical step in visual tracking. Recently, deep convolution network has demonstrated remarkable ability in visual tracking via leveraging robust high-level features. To obtain these high-level features, convolution and pooling operations are executed alternatively in deep convolution network. However, these operations lead to low spatial resolution feature maps which degrade the localization precision in tracking. While low level features have sufficient spatial resolution, their representation ability is insufficient. To mitigate this issue, we exploited deconvolution network in visual tracking. This deconvolution network works as a learnable upsampling layer which takes low-resolution high-level feature maps as input and outputs enlarged feature maps. Meanwhile, the low level feature maps are fused with these high level feature maps via a summarization operation to better represent target appearance. We formulate the network training as a regression issue and train this network end to end. Extensive experiments on two tracking benchmarks demonstrate the effectiveness of our method. |
topic |
Object tracking deep learning deconvolution neural network regression network |
url |
https://ieeexplore.ieee.org/document/8326476/ |
work_keys_str_mv |
AT xiankailu learningdeconvolutionalnetworkforobjecttracking AT honghuo learningdeconvolutionalnetworkforobjecttracking AT taofang learningdeconvolutionalnetworkforobjecttracking AT huanlongzhang learningdeconvolutionalnetworkforobjecttracking |
_version_ |
1724193784812535808 |