Adaptive Anchor Networks for Multi-Scale Object Detection in Remote Sensing Images
Accurate and effective object detection in remote sensing images plays an extremely important role in marine transport, environmental monitoring and military operations. Due to the powerful ability of feature representation, region-based convolutional neural networks (RCNNs) have been widely used in...
Main Authors: | , , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
IEEE
2020-01-01
|
Series: | IEEE Access |
Subjects: | |
Online Access: | https://ieeexplore.ieee.org/document/9044838/ |
id |
doaj-bdb5c4418fce4f44ae0568307841ff5e |
---|---|
record_format |
Article |
spelling |
doaj-bdb5c4418fce4f44ae0568307841ff5e2021-03-30T03:12:04ZengIEEEIEEE Access2169-35362020-01-018575525756510.1109/ACCESS.2020.29826589044838Adaptive Anchor Networks for Multi-Scale Object Detection in Remote Sensing ImagesMiaohui Zhang0Yunzhong Chen1https://orcid.org/0000-0002-2389-4188Xianxing Liu2Bingxue Lv3Jun Wang4Henan Key Laboratory of Big Data Analysis and Processing, Henan University, Kaifeng, ChinaHenan Key Laboratory of Big Data Analysis and Processing, Henan University, Kaifeng, ChinaHenan Key Laboratory of Big Data Analysis and Processing, Henan University, Kaifeng, ChinaCollege of Computer and Communication Engineering, Zhengzhou University of Light Industry, Zhengzhou, ChinaHenan Key Laboratory of Big Data Analysis and Processing, Henan University, Kaifeng, ChinaAccurate and effective object detection in remote sensing images plays an extremely important role in marine transport, environmental monitoring and military operations. Due to the powerful ability of feature representation, region-based convolutional neural networks (RCNNs) have been widely used in this field, which firstly generate candidate regions through extracted feature maps and then classify and locate objects. However, most of existing methods generally use traditional backbone networks to extract feature maps with a decreased spatial resolution because of the continuous down-sampling, which will weaken the information detected from small objects. Besides, sliding windows strategy is employed in these methods to generate fixed anchors with a preset scale on feature maps, which is inappropriate for multi-scale object detection in remote sensing images. To solve the above problems, a novel and effective object detection framework named DetNet-FPN (Feature Pyramid Network) is proposed in this paper, in which a feature pyramid with strong feature representation is created by combining feature maps of different spatial resolution, at the same time, the resolution of feature maps is maintained by involving dilation convolutions. Furthermore, to match the proposed backbone, the GA (Guided Anchoring)-RPN strategy is improved for adaptive anchor generation, this strategy simultaneously predicts the locations where the center of objects are likely to exist as well as the scales and aspect ratios at different locations. Extensive experiments and comprehensive evaluations demonstrate the effectiveness of the proposed framework on DOTA and NWPU VHR-10 datasets.https://ieeexplore.ieee.org/document/9044838/Convolutional neural networks (CNNs)object detectionremote sensing imagesmulti-scale feature fusionadaptive anchor |
collection |
DOAJ |
language |
English |
format |
Article |
sources |
DOAJ |
author |
Miaohui Zhang Yunzhong Chen Xianxing Liu Bingxue Lv Jun Wang |
spellingShingle |
Miaohui Zhang Yunzhong Chen Xianxing Liu Bingxue Lv Jun Wang Adaptive Anchor Networks for Multi-Scale Object Detection in Remote Sensing Images IEEE Access Convolutional neural networks (CNNs) object detection remote sensing images multi-scale feature fusion adaptive anchor |
author_facet |
Miaohui Zhang Yunzhong Chen Xianxing Liu Bingxue Lv Jun Wang |
author_sort |
Miaohui Zhang |
title |
Adaptive Anchor Networks for Multi-Scale Object Detection in Remote Sensing Images |
title_short |
Adaptive Anchor Networks for Multi-Scale Object Detection in Remote Sensing Images |
title_full |
Adaptive Anchor Networks for Multi-Scale Object Detection in Remote Sensing Images |
title_fullStr |
Adaptive Anchor Networks for Multi-Scale Object Detection in Remote Sensing Images |
title_full_unstemmed |
Adaptive Anchor Networks for Multi-Scale Object Detection in Remote Sensing Images |
title_sort |
adaptive anchor networks for multi-scale object detection in remote sensing images |
publisher |
IEEE |
series |
IEEE Access |
issn |
2169-3536 |
publishDate |
2020-01-01 |
description |
Accurate and effective object detection in remote sensing images plays an extremely important role in marine transport, environmental monitoring and military operations. Due to the powerful ability of feature representation, region-based convolutional neural networks (RCNNs) have been widely used in this field, which firstly generate candidate regions through extracted feature maps and then classify and locate objects. However, most of existing methods generally use traditional backbone networks to extract feature maps with a decreased spatial resolution because of the continuous down-sampling, which will weaken the information detected from small objects. Besides, sliding windows strategy is employed in these methods to generate fixed anchors with a preset scale on feature maps, which is inappropriate for multi-scale object detection in remote sensing images. To solve the above problems, a novel and effective object detection framework named DetNet-FPN (Feature Pyramid Network) is proposed in this paper, in which a feature pyramid with strong feature representation is created by combining feature maps of different spatial resolution, at the same time, the resolution of feature maps is maintained by involving dilation convolutions. Furthermore, to match the proposed backbone, the GA (Guided Anchoring)-RPN strategy is improved for adaptive anchor generation, this strategy simultaneously predicts the locations where the center of objects are likely to exist as well as the scales and aspect ratios at different locations. Extensive experiments and comprehensive evaluations demonstrate the effectiveness of the proposed framework on DOTA and NWPU VHR-10 datasets. |
topic |
Convolutional neural networks (CNNs) object detection remote sensing images multi-scale feature fusion adaptive anchor |
url |
https://ieeexplore.ieee.org/document/9044838/ |
work_keys_str_mv |
AT miaohuizhang adaptiveanchornetworksformultiscaleobjectdetectioninremotesensingimages AT yunzhongchen adaptiveanchornetworksformultiscaleobjectdetectioninremotesensingimages AT xianxingliu adaptiveanchornetworksformultiscaleobjectdetectioninremotesensingimages AT bingxuelv adaptiveanchornetworksformultiscaleobjectdetectioninremotesensingimages AT junwang adaptiveanchornetworksformultiscaleobjectdetectioninremotesensingimages |
_version_ |
1724183903224201216 |