Deep Learning for Feature Extraction in Remote Sensing: A Case-Study of Aerial Scene Classification
Scene classification relying on images is essential in many systems and applications related to remote sensing. The scientific interest in scene classification from remotely collected images is increasing, and many datasets and algorithms are being developed. The introduction of convolutional neural...
Main Authors: | , , , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
MDPI AG
2020-07-01
|
Series: | Sensors |
Subjects: | |
Online Access: | https://www.mdpi.com/1424-8220/20/14/3906 |
id |
doaj-0a530b871421422c959c7c46f7bfec05 |
---|---|
record_format |
Article |
spelling |
doaj-0a530b871421422c959c7c46f7bfec052020-11-25T02:53:12ZengMDPI AGSensors1424-82202020-07-01203906390610.3390/s20143906Deep Learning for Feature Extraction in Remote Sensing: A Case-Study of Aerial Scene ClassificationBiserka Petrovska0Eftim Zdravevski1Petre Lameski2Roberto Corizzo3Ivan Štajduhar4Jonatan Lerga5Ministry of Defense of Republic of North Macedonia, 1000 Skopje, North MacedoniaFaculty of Computer Science and Engineering, Saints Cyril and Methodius University, 1000 Skopje, North MacedoniaFaculty of Computer Science and Engineering, Saints Cyril and Methodius University, 1000 Skopje, North MacedoniaDepartment of Computer Science, University of Bari Aldo Moro, 70125 Bari, ItalyFaculty of Engineering, University of Rijeka, 51000 Rijeka, CroatiaFaculty of Engineering, University of Rijeka, 51000 Rijeka, CroatiaScene classification relying on images is essential in many systems and applications related to remote sensing. The scientific interest in scene classification from remotely collected images is increasing, and many datasets and algorithms are being developed. The introduction of convolutional neural networks (CNN) and other deep learning techniques contributed to vast improvements in the accuracy of image scene classification in such systems. To classify the scene from areal images, we used a two-stream deep architecture. We performed the first part of the classification, the feature extraction, using pre-trained CNN that extracts deep features of aerial images from different network layers: the average pooling layer or some of the previous convolutional layers. Next, we applied feature concatenation on extracted features from various neural networks, after dimensionality reduction was performed on enormous feature vectors. We experimented extensively with different CNN architectures, to get optimal results. Finally, we used the Support Vector Machine (SVM) for the classification of the concatenated features. The competitiveness of the examined technique was evaluated on two real-world datasets: UC Merced and WHU-RS. The obtained classification accuracies demonstrate that the considered method has competitive results compared to other cutting-edge techniques.https://www.mdpi.com/1424-8220/20/14/3906remote sensingconvolutional neural network (CNN)feature extractionfeature fusion |
collection |
DOAJ |
language |
English |
format |
Article |
sources |
DOAJ |
author |
Biserka Petrovska Eftim Zdravevski Petre Lameski Roberto Corizzo Ivan Štajduhar Jonatan Lerga |
spellingShingle |
Biserka Petrovska Eftim Zdravevski Petre Lameski Roberto Corizzo Ivan Štajduhar Jonatan Lerga Deep Learning for Feature Extraction in Remote Sensing: A Case-Study of Aerial Scene Classification Sensors remote sensing convolutional neural network (CNN) feature extraction feature fusion |
author_facet |
Biserka Petrovska Eftim Zdravevski Petre Lameski Roberto Corizzo Ivan Štajduhar Jonatan Lerga |
author_sort |
Biserka Petrovska |
title |
Deep Learning for Feature Extraction in Remote Sensing: A Case-Study of Aerial Scene Classification |
title_short |
Deep Learning for Feature Extraction in Remote Sensing: A Case-Study of Aerial Scene Classification |
title_full |
Deep Learning for Feature Extraction in Remote Sensing: A Case-Study of Aerial Scene Classification |
title_fullStr |
Deep Learning for Feature Extraction in Remote Sensing: A Case-Study of Aerial Scene Classification |
title_full_unstemmed |
Deep Learning for Feature Extraction in Remote Sensing: A Case-Study of Aerial Scene Classification |
title_sort |
deep learning for feature extraction in remote sensing: a case-study of aerial scene classification |
publisher |
MDPI AG |
series |
Sensors |
issn |
1424-8220 |
publishDate |
2020-07-01 |
description |
Scene classification relying on images is essential in many systems and applications related to remote sensing. The scientific interest in scene classification from remotely collected images is increasing, and many datasets and algorithms are being developed. The introduction of convolutional neural networks (CNN) and other deep learning techniques contributed to vast improvements in the accuracy of image scene classification in such systems. To classify the scene from areal images, we used a two-stream deep architecture. We performed the first part of the classification, the feature extraction, using pre-trained CNN that extracts deep features of aerial images from different network layers: the average pooling layer or some of the previous convolutional layers. Next, we applied feature concatenation on extracted features from various neural networks, after dimensionality reduction was performed on enormous feature vectors. We experimented extensively with different CNN architectures, to get optimal results. Finally, we used the Support Vector Machine (SVM) for the classification of the concatenated features. The competitiveness of the examined technique was evaluated on two real-world datasets: UC Merced and WHU-RS. The obtained classification accuracies demonstrate that the considered method has competitive results compared to other cutting-edge techniques. |
topic |
remote sensing convolutional neural network (CNN) feature extraction feature fusion |
url |
https://www.mdpi.com/1424-8220/20/14/3906 |
work_keys_str_mv |
AT biserkapetrovska deeplearningforfeatureextractioninremotesensingacasestudyofaerialsceneclassification AT eftimzdravevski deeplearningforfeatureextractioninremotesensingacasestudyofaerialsceneclassification AT petrelameski deeplearningforfeatureextractioninremotesensingacasestudyofaerialsceneclassification AT robertocorizzo deeplearningforfeatureextractioninremotesensingacasestudyofaerialsceneclassification AT ivanstajduhar deeplearningforfeatureextractioninremotesensingacasestudyofaerialsceneclassification AT jonatanlerga deeplearningforfeatureextractioninremotesensingacasestudyofaerialsceneclassification |
_version_ |
1724726170964983808 |