A survey on deep learning-based precise boundary recovery of semantic segmentation for images and point clouds
Precise localization of semantic segmentation is attracting increasing attention, and salient performances are dominated by deep learning-based methods, especially deep convolutional neural networks (DCNNs). However, the outputs from the final layer of DCNNs are not sufficiently localized for accura...
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2021-10-01
|
| Series: | International Journal of Applied Earth Observations and Geoinformation |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S0303243421001185 |
| _version_ | 1811228364638257152 |
|---|---|
| author | Rui Zhang Guangyun Li Thomas Wunderlich Li Wang |
| author_facet | Rui Zhang Guangyun Li Thomas Wunderlich Li Wang |
| author_sort | Rui Zhang |
| collection | DOAJ |
| description | Precise localization of semantic segmentation is attracting increasing attention, and salient performances are dominated by deep learning-based methods, especially deep convolutional neural networks (DCNNs). However, the outputs from the final layer of DCNNs are not sufficiently localized for accurate object boundaries due to their invariance properties, which makes precise boundary recovery of semantic segmentation an academically challenging question. Both 2D and 3D objects suffer from the same problem. Considering this, this paper conducts a comprehensive survey of precise boundary recovery for semantic segmentation, focusing mainly on 2D images and 3D point clouds. Firstly, we formulate the problem of potential boundary recovery for semantic segmentation based on DCNNs, elaborate on the terminology as well as background concepts in this field. Then, we categorize boundary recovery methods into four strategies according to their techniques and network architectures to discuss how they obtain accurate boundaries of semantic segmentation. Next, publicly available datasets on which they have been assessed are argued. To compare these datasets, we design diagrams based on five indicators to help researchers judge which are the ones that best suit their tasks. Moreover, we further compare and analyze the performance of all the reviewed methods through experimental results. Finally, current challenges and prospective research issues are discussed extensively. |
| first_indexed | 2024-04-12T09:57:42Z |
| format | Article |
| id | doaj.art-89659d0532c946ebb0e123cb7a62bdd1 |
| institution | Directory Open Access Journal |
| issn | 1569-8432 |
| language | English |
| last_indexed | 2024-04-12T09:57:42Z |
| publishDate | 2021-10-01 |
| publisher | Elsevier |
| record_format | Article |
| series | International Journal of Applied Earth Observations and Geoinformation |
| spelling | doaj.art-89659d0532c946ebb0e123cb7a62bdd12022-12-22T03:37:39ZengElsevierInternational Journal of Applied Earth Observations and Geoinformation1569-84322021-10-01102102411A survey on deep learning-based precise boundary recovery of semantic segmentation for images and point cloudsRui Zhang0Guangyun Li1Thomas Wunderlich2Li Wang3North China University of Water Resources and Electric Power, 450045 Zhengzhou, China; Chair of Geodesy, Technische Universität München, 80333 München, GermanyDepartment of Geospatial Information, PLA Information Engineering University, 450001 Zhengzhou, China; Corresponding author.Chair of Geodesy, Technische Universität München, 80333 München, GermanyDepartment of Geospatial Information, PLA Information Engineering University, 450001 Zhengzhou, ChinaPrecise localization of semantic segmentation is attracting increasing attention, and salient performances are dominated by deep learning-based methods, especially deep convolutional neural networks (DCNNs). However, the outputs from the final layer of DCNNs are not sufficiently localized for accurate object boundaries due to their invariance properties, which makes precise boundary recovery of semantic segmentation an academically challenging question. Both 2D and 3D objects suffer from the same problem. Considering this, this paper conducts a comprehensive survey of precise boundary recovery for semantic segmentation, focusing mainly on 2D images and 3D point clouds. Firstly, we formulate the problem of potential boundary recovery for semantic segmentation based on DCNNs, elaborate on the terminology as well as background concepts in this field. Then, we categorize boundary recovery methods into four strategies according to their techniques and network architectures to discuss how they obtain accurate boundaries of semantic segmentation. Next, publicly available datasets on which they have been assessed are argued. To compare these datasets, we design diagrams based on five indicators to help researchers judge which are the ones that best suit their tasks. Moreover, we further compare and analyze the performance of all the reviewed methods through experimental results. Finally, current challenges and prospective research issues are discussed extensively.http://www.sciencedirect.com/science/article/pii/S0303243421001185Precise boundary recoverySemantic segmentationDCNNs2D images3D point clouds |
| spellingShingle | Rui Zhang Guangyun Li Thomas Wunderlich Li Wang A survey on deep learning-based precise boundary recovery of semantic segmentation for images and point clouds International Journal of Applied Earth Observations and Geoinformation Precise boundary recovery Semantic segmentation DCNNs 2D images 3D point clouds |
| title | A survey on deep learning-based precise boundary recovery of semantic segmentation for images and point clouds |
| title_full | A survey on deep learning-based precise boundary recovery of semantic segmentation for images and point clouds |
| title_fullStr | A survey on deep learning-based precise boundary recovery of semantic segmentation for images and point clouds |
| title_full_unstemmed | A survey on deep learning-based precise boundary recovery of semantic segmentation for images and point clouds |
| title_short | A survey on deep learning-based precise boundary recovery of semantic segmentation for images and point clouds |
| title_sort | survey on deep learning based precise boundary recovery of semantic segmentation for images and point clouds |
| topic | Precise boundary recovery Semantic segmentation DCNNs 2D images 3D point clouds |
| url | http://www.sciencedirect.com/science/article/pii/S0303243421001185 |
| work_keys_str_mv | AT ruizhang asurveyondeeplearningbasedpreciseboundaryrecoveryofsemanticsegmentationforimagesandpointclouds AT guangyunli asurveyondeeplearningbasedpreciseboundaryrecoveryofsemanticsegmentationforimagesandpointclouds AT thomaswunderlich asurveyondeeplearningbasedpreciseboundaryrecoveryofsemanticsegmentationforimagesandpointclouds AT liwang asurveyondeeplearningbasedpreciseboundaryrecoveryofsemanticsegmentationforimagesandpointclouds AT ruizhang surveyondeeplearningbasedpreciseboundaryrecoveryofsemanticsegmentationforimagesandpointclouds AT guangyunli surveyondeeplearningbasedpreciseboundaryrecoveryofsemanticsegmentationforimagesandpointclouds AT thomaswunderlich surveyondeeplearningbasedpreciseboundaryrecoveryofsemanticsegmentationforimagesandpointclouds AT liwang surveyondeeplearningbasedpreciseboundaryrecoveryofsemanticsegmentationforimagesandpointclouds |