Applying Fully Convolutional Architectures for Semantic Segmentation of a Single Tree Species in Urban Environment on High Resolution UAV Optical Imagery
This study proposes and evaluates five deep fully convolutional networks (FCNs) for the semantic segmentation of a single tree species: SegNet, U-Net, FC-DenseNet, and two DeepLabv3+ variants. The performance of the FCN designs is evaluated experimentally in terms of classification accuracy and comp...
| Main Authors: | , , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2020-01-01
|
| Series: | Sensors |
| Subjects: | |
| Online Access: | https://www.mdpi.com/1424-8220/20/2/563 |
| _version_ | 1798024186994622464 |
|---|---|
| author | Daliana Lobo Torres Raul Queiroz Feitosa Patrick Nigri Happ Laura Elena Cué La Rosa José Marcato Junior José Martins Patrik Olã Bressan Wesley Nunes Gonçalves Veraldo Liesenberg |
| author_facet | Daliana Lobo Torres Raul Queiroz Feitosa Patrick Nigri Happ Laura Elena Cué La Rosa José Marcato Junior José Martins Patrik Olã Bressan Wesley Nunes Gonçalves Veraldo Liesenberg |
| author_sort | Daliana Lobo Torres |
| collection | DOAJ |
| description | This study proposes and evaluates five deep fully convolutional networks (FCNs) for the semantic segmentation of a single tree species: SegNet, U-Net, FC-DenseNet, and two DeepLabv3+ variants. The performance of the FCN designs is evaluated experimentally in terms of classification accuracy and computational load. We also verify the benefits of fully connected conditional random fields (CRFs) as a post-processing step to improve the segmentation maps. The analysis is conducted on a set of images captured by an RGB camera aboard a UAV flying over an urban area. The dataset also contains a mask that indicates the occurrence of an endangered species called <i>Dipteryx alata</i> Vogel, also known as cumbaru, taken as the species to be identified. The experimental analysis shows the effectiveness of each design and reports average overall accuracy ranging from 88.9% to 96.7%, an F1-score between 87.0% and 96.1%, and IoU from 77.1% to 92.5%. We also realize that CRF consistently improves the performance, but at a high computational cost. |
| first_indexed | 2024-04-11T17:58:24Z |
| format | Article |
| id | doaj.art-37f1de7a47da4f8bb9924c450f5e1a91 |
| institution | Directory Open Access Journal |
| issn | 1424-8220 |
| language | English |
| last_indexed | 2024-04-11T17:58:24Z |
| publishDate | 2020-01-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Sensors |
| spelling | doaj.art-37f1de7a47da4f8bb9924c450f5e1a912022-12-22T04:10:35ZengMDPI AGSensors1424-82202020-01-0120256310.3390/s20020563s20020563Applying Fully Convolutional Architectures for Semantic Segmentation of a Single Tree Species in Urban Environment on High Resolution UAV Optical ImageryDaliana Lobo Torres0Raul Queiroz Feitosa1Patrick Nigri Happ2Laura Elena Cué La Rosa3José Marcato Junior4José Martins5Patrik Olã Bressan6Wesley Nunes Gonçalves7Veraldo Liesenberg8Department of Electrical Engineering, Pontifical Catholic University of Rio de Janeiro, Rio de Janeiro 22451-900, BrazilDepartment of Electrical Engineering, Pontifical Catholic University of Rio de Janeiro, Rio de Janeiro 22451-900, BrazilDepartment of Electrical Engineering, Pontifical Catholic University of Rio de Janeiro, Rio de Janeiro 22451-900, BrazilDepartment of Electrical Engineering, Pontifical Catholic University of Rio de Janeiro, Rio de Janeiro 22451-900, BrazilFaculty of Engineering, Architecture and Urbanism and Geography, Federal University of Mato Grosso do Sul, Campo Grande 79070-900, BrazilFaculty of Engineering, Architecture and Urbanism and Geography, Federal University of Mato Grosso do Sul, Campo Grande 79070-900, BrazilFederal Institute of Mato Grosso do Sul, Jardim 79240-000, BrazilFaculty of Engineering, Architecture and Urbanism and Geography, Federal University of Mato Grosso do Sul, Campo Grande 79070-900, BrazilDepartment of Forest Engineering, Santa Catarina State University, Lages 88520-000, BrazilThis study proposes and evaluates five deep fully convolutional networks (FCNs) for the semantic segmentation of a single tree species: SegNet, U-Net, FC-DenseNet, and two DeepLabv3+ variants. The performance of the FCN designs is evaluated experimentally in terms of classification accuracy and computational load. We also verify the benefits of fully connected conditional random fields (CRFs) as a post-processing step to improve the segmentation maps. The analysis is conducted on a set of images captured by an RGB camera aboard a UAV flying over an urban area. The dataset also contains a mask that indicates the occurrence of an endangered species called <i>Dipteryx alata</i> Vogel, also known as cumbaru, taken as the species to be identified. The experimental analysis shows the effectiveness of each design and reports average overall accuracy ranging from 88.9% to 96.7%, an F1-score between 87.0% and 96.1%, and IoU from 77.1% to 92.5%. We also realize that CRF consistently improves the performance, but at a high computational cost.https://www.mdpi.com/1424-8220/20/2/563deep learningfully convolution neural networkssemantic segmentationunmanned aerial vehicle (uav) |
| spellingShingle | Daliana Lobo Torres Raul Queiroz Feitosa Patrick Nigri Happ Laura Elena Cué La Rosa José Marcato Junior José Martins Patrik Olã Bressan Wesley Nunes Gonçalves Veraldo Liesenberg Applying Fully Convolutional Architectures for Semantic Segmentation of a Single Tree Species in Urban Environment on High Resolution UAV Optical Imagery Sensors deep learning fully convolution neural networks semantic segmentation unmanned aerial vehicle (uav) |
| title | Applying Fully Convolutional Architectures for Semantic Segmentation of a Single Tree Species in Urban Environment on High Resolution UAV Optical Imagery |
| title_full | Applying Fully Convolutional Architectures for Semantic Segmentation of a Single Tree Species in Urban Environment on High Resolution UAV Optical Imagery |
| title_fullStr | Applying Fully Convolutional Architectures for Semantic Segmentation of a Single Tree Species in Urban Environment on High Resolution UAV Optical Imagery |
| title_full_unstemmed | Applying Fully Convolutional Architectures for Semantic Segmentation of a Single Tree Species in Urban Environment on High Resolution UAV Optical Imagery |
| title_short | Applying Fully Convolutional Architectures for Semantic Segmentation of a Single Tree Species in Urban Environment on High Resolution UAV Optical Imagery |
| title_sort | applying fully convolutional architectures for semantic segmentation of a single tree species in urban environment on high resolution uav optical imagery |
| topic | deep learning fully convolution neural networks semantic segmentation unmanned aerial vehicle (uav) |
| url | https://www.mdpi.com/1424-8220/20/2/563 |
| work_keys_str_mv | AT dalianalobotorres applyingfullyconvolutionalarchitecturesforsemanticsegmentationofasingletreespeciesinurbanenvironmentonhighresolutionuavopticalimagery AT raulqueirozfeitosa applyingfullyconvolutionalarchitecturesforsemanticsegmentationofasingletreespeciesinurbanenvironmentonhighresolutionuavopticalimagery AT patricknigrihapp applyingfullyconvolutionalarchitecturesforsemanticsegmentationofasingletreespeciesinurbanenvironmentonhighresolutionuavopticalimagery AT lauraelenacuelarosa applyingfullyconvolutionalarchitecturesforsemanticsegmentationofasingletreespeciesinurbanenvironmentonhighresolutionuavopticalimagery AT josemarcatojunior applyingfullyconvolutionalarchitecturesforsemanticsegmentationofasingletreespeciesinurbanenvironmentonhighresolutionuavopticalimagery AT josemartins applyingfullyconvolutionalarchitecturesforsemanticsegmentationofasingletreespeciesinurbanenvironmentonhighresolutionuavopticalimagery AT patrikolabressan applyingfullyconvolutionalarchitecturesforsemanticsegmentationofasingletreespeciesinurbanenvironmentonhighresolutionuavopticalimagery AT wesleynunesgoncalves applyingfullyconvolutionalarchitecturesforsemanticsegmentationofasingletreespeciesinurbanenvironmentonhighresolutionuavopticalimagery AT veraldoliesenberg applyingfullyconvolutionalarchitecturesforsemanticsegmentationofasingletreespeciesinurbanenvironmentonhighresolutionuavopticalimagery |