Classification of Very-High-Spatial-Resolution Aerial Images Based on Multiscale Features with Limited Semantic Information
Recently, deep learning has become the most innovative trend for a variety of high-spatial-resolution remote sensing imaging applications. However, large-scale land cover classification via traditional convolutional neural networks (CNNs) with sliding windows is computationally expensive and produce...
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2021-01-01
|
| Series: | Remote Sensing |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2072-4292/13/3/364 |
| _version_ | 1797408720524673024 |
|---|---|
| author | Han Gao Jinhui Guo Peng Guo Xiuwan Chen |
| author_facet | Han Gao Jinhui Guo Peng Guo Xiuwan Chen |
| author_sort | Han Gao |
| collection | DOAJ |
| description | Recently, deep learning has become the most innovative trend for a variety of high-spatial-resolution remote sensing imaging applications. However, large-scale land cover classification via traditional convolutional neural networks (CNNs) with sliding windows is computationally expensive and produces coarse results. Additionally, although such supervised learning approaches have performed well, collecting and annotating datasets for every task are extremely laborious, especially for those fully supervised cases where the pixel-level ground-truth labels are dense. In this work, we propose a new object-oriented deep learning framework that leverages residual networks with different depths to learn adjacent feature representations by embedding a multibranch architecture in the deep learning pipeline. The idea is to exploit limited training data at different neighboring scales to make a tradeoff between weak semantics and strong feature representations for operational land cover mapping tasks. We draw from established geographic object-based image analysis (GEOBIA) as an auxiliary module to reduce the computational burden of spatial reasoning and optimize the classification boundaries. We evaluated the proposed approach on two subdecimeter-resolution datasets involving both urban and rural landscapes. It presented better classification accuracy (88.9%) compared to traditional object-based deep learning methods and achieves an excellent inference time (11.3 s/ha). |
| first_indexed | 2024-03-09T04:02:36Z |
| format | Article |
| id | doaj.art-7ff8e80b0a5141e3ab25ce9505b216c9 |
| institution | Directory Open Access Journal |
| issn | 2072-4292 |
| language | English |
| last_indexed | 2024-03-09T04:02:36Z |
| publishDate | 2021-01-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Remote Sensing |
| spelling | doaj.art-7ff8e80b0a5141e3ab25ce9505b216c92023-12-03T14:10:59ZengMDPI AGRemote Sensing2072-42922021-01-0113336410.3390/rs13030364Classification of Very-High-Spatial-Resolution Aerial Images Based on Multiscale Features with Limited Semantic InformationHan Gao0Jinhui Guo1Peng Guo2Xiuwan Chen3Institute of Remote Sensing and Geographic Information System, Peking University, Beijing 100871, ChinaInstitute of Remote Sensing and Geographic Information System, Peking University, Beijing 100871, ChinaInstitute of Remote Sensing and Geographic Information System, Peking University, Beijing 100871, ChinaInstitute of Remote Sensing and Geographic Information System, Peking University, Beijing 100871, ChinaRecently, deep learning has become the most innovative trend for a variety of high-spatial-resolution remote sensing imaging applications. However, large-scale land cover classification via traditional convolutional neural networks (CNNs) with sliding windows is computationally expensive and produces coarse results. Additionally, although such supervised learning approaches have performed well, collecting and annotating datasets for every task are extremely laborious, especially for those fully supervised cases where the pixel-level ground-truth labels are dense. In this work, we propose a new object-oriented deep learning framework that leverages residual networks with different depths to learn adjacent feature representations by embedding a multibranch architecture in the deep learning pipeline. The idea is to exploit limited training data at different neighboring scales to make a tradeoff between weak semantics and strong feature representations for operational land cover mapping tasks. We draw from established geographic object-based image analysis (GEOBIA) as an auxiliary module to reduce the computational burden of spatial reasoning and optimize the classification boundaries. We evaluated the proposed approach on two subdecimeter-resolution datasets involving both urban and rural landscapes. It presented better classification accuracy (88.9%) compared to traditional object-based deep learning methods and achieves an excellent inference time (11.3 s/ha).https://www.mdpi.com/2072-4292/13/3/364deep learningaerial imageryconvolutional neural networkobject-based classification |
| spellingShingle | Han Gao Jinhui Guo Peng Guo Xiuwan Chen Classification of Very-High-Spatial-Resolution Aerial Images Based on Multiscale Features with Limited Semantic Information Remote Sensing deep learning aerial imagery convolutional neural network object-based classification |
| title | Classification of Very-High-Spatial-Resolution Aerial Images Based on Multiscale Features with Limited Semantic Information |
| title_full | Classification of Very-High-Spatial-Resolution Aerial Images Based on Multiscale Features with Limited Semantic Information |
| title_fullStr | Classification of Very-High-Spatial-Resolution Aerial Images Based on Multiscale Features with Limited Semantic Information |
| title_full_unstemmed | Classification of Very-High-Spatial-Resolution Aerial Images Based on Multiscale Features with Limited Semantic Information |
| title_short | Classification of Very-High-Spatial-Resolution Aerial Images Based on Multiscale Features with Limited Semantic Information |
| title_sort | classification of very high spatial resolution aerial images based on multiscale features with limited semantic information |
| topic | deep learning aerial imagery convolutional neural network object-based classification |
| url | https://www.mdpi.com/2072-4292/13/3/364 |
| work_keys_str_mv | AT hangao classificationofveryhighspatialresolutionaerialimagesbasedonmultiscalefeatureswithlimitedsemanticinformation AT jinhuiguo classificationofveryhighspatialresolutionaerialimagesbasedonmultiscalefeatureswithlimitedsemanticinformation AT pengguo classificationofveryhighspatialresolutionaerialimagesbasedonmultiscalefeatureswithlimitedsemanticinformation AT xiuwanchen classificationofveryhighspatialresolutionaerialimagesbasedonmultiscalefeatureswithlimitedsemanticinformation |