TU Darmstadt / ULB / TUbiblio

A Light and Faster Regional Convolutional Neural Network for Object Detection in Optical Remote Sensing Images

Ding, Peng and Zhang, Ye and Deng, Wei-Jian and Jia, Ping and Kuijper, Arjan (2018):
A Light and Faster Regional Convolutional Neural Network for Object Detection in Optical Remote Sensing Images.
141, In: ISPRS Journal of Photogrammetry and Remote Sensing, pp. 208-218, ISSN 09242716, DOI: 10.1016/j.isprsjprs.2018.05.005,
[Online-Edition: https://doi.org/10.1016/j.isprsjprs.2018.05.005],
[Article]

Abstract

Detection of objects from satellite optical remote sensing images is very important for many commercial and governmental applications. With the development of deep convolutional neural networks (deep CNNs), the field of object detection has seen tremendous advances. Currently, objects in satellite remote sensing images can be detected using deep CNNs. In general, optical remote sensing images contain many dense and small objects, and the use of the original Faster Regional CNN framework does not yield a suitably high precision. Therefore, after careful analysis we adopt dense convoluted networks, a multi-scale representation and various combinations of improvement schemes to enhance the structure of the base VGG16-Net for improving the precision. We propose an approach to reduce the test-time (detection time) and memory requirements. To validate the effectiveness of our approach, we perform experiments using satellite remote sensing image datasets of aircraft and automobiles. The results show that the improved network structure can detect objects in satellite optical remote sensing images more accurately and efficiently.

Item Type: Article
Erschienen: 2018
Creators: Ding, Peng and Zhang, Ye and Deng, Wei-Jian and Jia, Ping and Kuijper, Arjan
Title: A Light and Faster Regional Convolutional Neural Network for Object Detection in Optical Remote Sensing Images
Language: English
Abstract:

Detection of objects from satellite optical remote sensing images is very important for many commercial and governmental applications. With the development of deep convolutional neural networks (deep CNNs), the field of object detection has seen tremendous advances. Currently, objects in satellite remote sensing images can be detected using deep CNNs. In general, optical remote sensing images contain many dense and small objects, and the use of the original Faster Regional CNN framework does not yield a suitably high precision. Therefore, after careful analysis we adopt dense convoluted networks, a multi-scale representation and various combinations of improvement schemes to enhance the structure of the base VGG16-Net for improving the precision. We propose an approach to reduce the test-time (detection time) and memory requirements. To validate the effectiveness of our approach, we perform experiments using satellite remote sensing image datasets of aircraft and automobiles. The results show that the improved network structure can detect objects in satellite optical remote sensing images more accurately and efficiently.

Journal or Publication Title: ISPRS Journal of Photogrammetry and Remote Sensing
Volume: 141
Uncontrolled Keywords: Convolutional Neural Networks (CNN), Deep learning, Object detection
Divisions: 20 Department of Computer Science
20 Department of Computer Science > Mathematical and Applied Visual Computing
Date Deposited: 19 Jun 2019 11:18
DOI: 10.1016/j.isprsjprs.2018.05.005
Official URL: https://doi.org/10.1016/j.isprsjprs.2018.05.005
Export:
Suche nach Titel in: TUfind oder in Google
Send an inquiry Send an inquiry

Options (only for editors)

View Item View Item