Foreground Feature Enhancement for Object Detection
Deep convolutional neural networks have shown great success in object detection. Most object detection methods focus on improving network architecture and introducing additional objective functions to improve the discrimination of object detectors, while the informative annotations of the training d...
Main Authors: | , , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
IEEE
2019-01-01
|
Series: | IEEE Access |
Subjects: | |
Online Access: | https://ieeexplore.ieee.org/document/8684952/ |
id |
doaj-9748855a6efc4e4295b8e77b6915aa00 |
---|---|
record_format |
Article |
spelling |
doaj-9748855a6efc4e4295b8e77b6915aa002021-03-29T22:15:38ZengIEEEIEEE Access2169-35362019-01-017492234923110.1109/ACCESS.2019.29086308684952Foreground Feature Enhancement for Object DetectionShenwang Jiang0https://orcid.org/0000-0002-0914-4954Tingfa Xu1Jianan Li2Ziyi Shen3Jie Guo4Beijing Institute of Technology, Beijing, ChinaBeijing Institute of Technology, Beijing, ChinaBeijing Institute of Technology, Beijing, ChinaBeijing Institute of Technology, Beijing, ChinaBeijing Institute of Technology, Beijing, ChinaDeep convolutional neural networks have shown great success in object detection. Most object detection methods focus on improving network architecture and introducing additional objective functions to improve the discrimination of object detectors, while the informative annotations of the training data obtained from enormous human effort are mainly used in the last stage of the network for producing supervisions, thus being under-explored. In this paper, we propose to take further advantage of bounding box annotations to highlight the feature map of foreground objects by erasing background noise with a novel Mask loss, in which process L<sub>2</sub> norm is further incorporated to avoid degenerated features. The extensive experiments on PASCAL VOC 2007, VOC 2012, and COCO 2017 will demonstrate the proposed method can greatly improve detection performance compared with baseline models, thus achieving competitive results.https://ieeexplore.ieee.org/document/8684952/Object detectionfeature enhancementdeep learning |
collection |
DOAJ |
language |
English |
format |
Article |
sources |
DOAJ |
author |
Shenwang Jiang Tingfa Xu Jianan Li Ziyi Shen Jie Guo |
spellingShingle |
Shenwang Jiang Tingfa Xu Jianan Li Ziyi Shen Jie Guo Foreground Feature Enhancement for Object Detection IEEE Access Object detection feature enhancement deep learning |
author_facet |
Shenwang Jiang Tingfa Xu Jianan Li Ziyi Shen Jie Guo |
author_sort |
Shenwang Jiang |
title |
Foreground Feature Enhancement for Object Detection |
title_short |
Foreground Feature Enhancement for Object Detection |
title_full |
Foreground Feature Enhancement for Object Detection |
title_fullStr |
Foreground Feature Enhancement for Object Detection |
title_full_unstemmed |
Foreground Feature Enhancement for Object Detection |
title_sort |
foreground feature enhancement for object detection |
publisher |
IEEE |
series |
IEEE Access |
issn |
2169-3536 |
publishDate |
2019-01-01 |
description |
Deep convolutional neural networks have shown great success in object detection. Most object detection methods focus on improving network architecture and introducing additional objective functions to improve the discrimination of object detectors, while the informative annotations of the training data obtained from enormous human effort are mainly used in the last stage of the network for producing supervisions, thus being under-explored. In this paper, we propose to take further advantage of bounding box annotations to highlight the feature map of foreground objects by erasing background noise with a novel Mask loss, in which process L<sub>2</sub> norm is further incorporated to avoid degenerated features. The extensive experiments on PASCAL VOC 2007, VOC 2012, and COCO 2017 will demonstrate the proposed method can greatly improve detection performance compared with baseline models, thus achieving competitive results. |
topic |
Object detection feature enhancement deep learning |
url |
https://ieeexplore.ieee.org/document/8684952/ |
work_keys_str_mv |
AT shenwangjiang foregroundfeatureenhancementforobjectdetection AT tingfaxu foregroundfeatureenhancementforobjectdetection AT jiananli foregroundfeatureenhancementforobjectdetection AT ziyishen foregroundfeatureenhancementforobjectdetection AT jieguo foregroundfeatureenhancementforobjectdetection |
_version_ |
1724191983797272576 |