DGANet: Dynamic Gradient Adjustment Anchor-Free Object Detection in Optical Remote Sensing Images

Remote sensing image object detection has been studied by many researchers in recent years using deep neural networks. However, optical remote sensing images contain many scenes with small and dense objects, resulting in a high rate of misrecognition. Firstly, in this work we selected a deep layer a...

Full description

Bibliographic Details
Main Authors: Peng Wang, Yanxiong Niu, Rui Xiong, Fu Ma, Chunxi Zhang
Format: Article
Language:English
Published: MDPI AG 2021-04-01
Series:Remote Sensing
Subjects:
Online Access:https://www.mdpi.com/2072-4292/13/9/1642
_version_ 1797536688522657792
author Peng Wang
Yanxiong Niu
Rui Xiong
Fu Ma
Chunxi Zhang
author_facet Peng Wang
Yanxiong Niu
Rui Xiong
Fu Ma
Chunxi Zhang
author_sort Peng Wang
collection DOAJ
description Remote sensing image object detection has been studied by many researchers in recent years using deep neural networks. However, optical remote sensing images contain many scenes with small and dense objects, resulting in a high rate of misrecognition. Firstly, in this work we selected a deep layer aggregation network with updated deformable convolution layers as the backbone to extract object features. The detection and classification of objects was based on the center-point network without non-maximum suppression. Secondly, the dynamic gradient adjustment embedded into the classification loss function was put forward to harmonize the quantity imbalance between easy and hard examples, as well as between positive and negative examples. Furthermore, the complete intersection over union (CIoU) loss function was selected as the objective function of bounding box regression, which achieves better convergence speed and accuracy. Finally, in order to validate the effectiveness and precision of the dynamic gradient adjustment network (DGANet), we conducted a series of experiments in remote sensing public datasets UCAS-AOD and LEVIR. The comparison experiments demonstrate that the DGANet achieves a more accurate detection result in optical remote sensing images.
first_indexed 2024-03-10T12:04:20Z
format Article
id doaj.art-f44c0f614af54fa093c8bd913a8404c5
institution Directory Open Access Journal
issn 2072-4292
language English
last_indexed 2024-03-10T12:04:20Z
publishDate 2021-04-01
publisher MDPI AG
record_format Article
series Remote Sensing
spelling doaj.art-f44c0f614af54fa093c8bd913a8404c52023-11-21T16:43:51ZengMDPI AGRemote Sensing2072-42922021-04-01139164210.3390/rs13091642DGANet: Dynamic Gradient Adjustment Anchor-Free Object Detection in Optical Remote Sensing ImagesPeng Wang0Yanxiong Niu1Rui Xiong2Fu Ma3Chunxi Zhang4School of Instrumentation and Optoelectronic Engineering, Beihang University, Beijing 100191, ChinaSchool of Instrumentation and Optoelectronic Engineering, Beihang University, Beijing 100191, ChinaSchool of Instrumentation and Optoelectronic Engineering, Beihang University, Beijing 100191, ChinaSchool of Instrumentation and Optoelectronic Engineering, Beihang University, Beijing 100191, ChinaSchool of Instrumentation and Optoelectronic Engineering, Beihang University, Beijing 100191, ChinaRemote sensing image object detection has been studied by many researchers in recent years using deep neural networks. However, optical remote sensing images contain many scenes with small and dense objects, resulting in a high rate of misrecognition. Firstly, in this work we selected a deep layer aggregation network with updated deformable convolution layers as the backbone to extract object features. The detection and classification of objects was based on the center-point network without non-maximum suppression. Secondly, the dynamic gradient adjustment embedded into the classification loss function was put forward to harmonize the quantity imbalance between easy and hard examples, as well as between positive and negative examples. Furthermore, the complete intersection over union (CIoU) loss function was selected as the objective function of bounding box regression, which achieves better convergence speed and accuracy. Finally, in order to validate the effectiveness and precision of the dynamic gradient adjustment network (DGANet), we conducted a series of experiments in remote sensing public datasets UCAS-AOD and LEVIR. The comparison experiments demonstrate that the DGANet achieves a more accurate detection result in optical remote sensing images.https://www.mdpi.com/2072-4292/13/9/1642deep learningobject detectionremote sensing imagescenter-point network
spellingShingle Peng Wang
Yanxiong Niu
Rui Xiong
Fu Ma
Chunxi Zhang
DGANet: Dynamic Gradient Adjustment Anchor-Free Object Detection in Optical Remote Sensing Images
Remote Sensing
deep learning
object detection
remote sensing images
center-point network
title DGANet: Dynamic Gradient Adjustment Anchor-Free Object Detection in Optical Remote Sensing Images
title_full DGANet: Dynamic Gradient Adjustment Anchor-Free Object Detection in Optical Remote Sensing Images
title_fullStr DGANet: Dynamic Gradient Adjustment Anchor-Free Object Detection in Optical Remote Sensing Images
title_full_unstemmed DGANet: Dynamic Gradient Adjustment Anchor-Free Object Detection in Optical Remote Sensing Images
title_short DGANet: Dynamic Gradient Adjustment Anchor-Free Object Detection in Optical Remote Sensing Images
title_sort dganet dynamic gradient adjustment anchor free object detection in optical remote sensing images
topic deep learning
object detection
remote sensing images
center-point network
url https://www.mdpi.com/2072-4292/13/9/1642
work_keys_str_mv AT pengwang dganetdynamicgradientadjustmentanchorfreeobjectdetectioninopticalremotesensingimages
AT yanxiongniu dganetdynamicgradientadjustmentanchorfreeobjectdetectioninopticalremotesensingimages
AT ruixiong dganetdynamicgradientadjustmentanchorfreeobjectdetectioninopticalremotesensingimages
AT fuma dganetdynamicgradientadjustmentanchorfreeobjectdetectioninopticalremotesensingimages
AT chunxizhang dganetdynamicgradientadjustmentanchorfreeobjectdetectioninopticalremotesensingimages