DBCG-Net: Dual Branch Calibration Guided Deep Network for UAV Images Semantic Segmentation
Unmanned aerial vehicle (UAV) remote sensing images used for semantic segmentation possess distinct features compared to urban street scene images, including high resolution and a complex background. Spatial information plays a pivotal role in enhancing the performance of semantic segmentation for h...
| Main Authors: | , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2024-01-01
|
| Series: | IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/10475391/ |
| _version_ | 1797206098455822336 |
|---|---|
| author | Chaoyun Mai Yibo Wu Yikui Zhai Hao Quan Jianhong Zhou Angelo Genovese Vincenzo Piuri Fabio Scotti |
| author_facet | Chaoyun Mai Yibo Wu Yikui Zhai Hao Quan Jianhong Zhou Angelo Genovese Vincenzo Piuri Fabio Scotti |
| author_sort | Chaoyun Mai |
| collection | DOAJ |
| description | Unmanned aerial vehicle (UAV) remote sensing images used for semantic segmentation possess distinct features compared to urban street scene images, including high resolution and a complex background. Spatial information plays a pivotal role in enhancing the performance of semantic segmentation for high-resolution images. The dual-branch architecture for semantic segmentation incorporates supplementary branches to capture spatial information. However, prior research on dual-branch semantic segmentation neglected the interaction between the contextual and spatial branches, leading to suboptimal model performance. In this discourse, the article introduces a dual-branch semantic segmentation framework. This design advances the system's understanding of spatial information while facilitating inter-branch learning through two key modules. Initially, the spatial calibration feature extraction module employs frequency domain processing and learning tactics distinct from the contextual approach to generate image features under varied noise conditions. Calibration is achieved by generating features from diverse angles. Subsequently, the spatially-guided loss function directs the acquisition of spatial information for the spatial branch by condensing the deep image characteristics for the context branch. To assess the generalization capacity of the proposed method, experiments will be conducted on three different datasets. The proposed method's modules will be integrated into three representative dual-branch networks, allowing assessment of the generalization capacity of the key DBCG components. Empirical evidence demonstrates that this approach is highly effective, significantly surpassing the performance of the baseline network. |
| first_indexed | 2024-04-24T09:01:37Z |
| format | Article |
| id | doaj.art-32cc3d9421ad49148ab5ad7baefc310d |
| institution | Directory Open Access Journal |
| issn | 1939-1404 2151-1535 |
| language | English |
| last_indexed | 2024-04-24T09:01:37Z |
| publishDate | 2024-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing |
| spelling | doaj.art-32cc3d9421ad49148ab5ad7baefc310d2024-04-15T23:00:12ZengIEEEIEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing1939-14042151-15352024-01-01177932794510.1109/JSTARS.2024.337869510475391DBCG-Net: Dual Branch Calibration Guided Deep Network for UAV Images Semantic SegmentationChaoyun Mai0https://orcid.org/0000-0003-4353-5868Yibo Wu1https://orcid.org/0000-0002-7802-4913Yikui Zhai2https://orcid.org/0000-0003-0154-9743Hao Quan3https://orcid.org/0000-0002-1107-0069Jianhong Zhou4https://orcid.org/0009-0001-6091-7318Angelo Genovese5https://orcid.org/0000-0002-3683-4723Vincenzo Piuri6https://orcid.org/0000-0003-3178-8198Fabio Scotti7https://orcid.org/0000-0002-4277-3701School of Electronics and Information Engineering, Wuyi University, Jiangmen, Guangdong, ChinaSchool of Electronics and Information Engineering, Wuyi University, Jiangmen, Guangdong, ChinaSchool of Electronics and Information Engineering, Wuyi University, Jiangmen, Guangdong, ChinaDipartimento di Elettronica, Informazione e Bioingegneria, Politecnico di Milano, Milano, ItalySchool of Electronics and Information Engineering, Wuyi University, Jiangmen, Guangdong, ChinaDipartimento di Informatica, Universitá Degli Studi di Milano, Milano, ItalyDipartimento di Informatica, Universitá Degli Studi di Milano, Milano, ItalyDipartimento di Informatica, Universitá Degli Studi di Milano, Milano, ItalyUnmanned aerial vehicle (UAV) remote sensing images used for semantic segmentation possess distinct features compared to urban street scene images, including high resolution and a complex background. Spatial information plays a pivotal role in enhancing the performance of semantic segmentation for high-resolution images. The dual-branch architecture for semantic segmentation incorporates supplementary branches to capture spatial information. However, prior research on dual-branch semantic segmentation neglected the interaction between the contextual and spatial branches, leading to suboptimal model performance. In this discourse, the article introduces a dual-branch semantic segmentation framework. This design advances the system's understanding of spatial information while facilitating inter-branch learning through two key modules. Initially, the spatial calibration feature extraction module employs frequency domain processing and learning tactics distinct from the contextual approach to generate image features under varied noise conditions. Calibration is achieved by generating features from diverse angles. Subsequently, the spatially-guided loss function directs the acquisition of spatial information for the spatial branch by condensing the deep image characteristics for the context branch. To assess the generalization capacity of the proposed method, experiments will be conducted on three different datasets. The proposed method's modules will be integrated into three representative dual-branch networks, allowing assessment of the generalization capacity of the key DBCG components. Empirical evidence demonstrates that this approach is highly effective, significantly surpassing the performance of the baseline network.https://ieeexplore.ieee.org/document/10475391/Convolutional neural network (CNN)deep learningdual-branch calibration guided networksemantic segmentationunmanned aerial vehicles (UAVs) |
| spellingShingle | Chaoyun Mai Yibo Wu Yikui Zhai Hao Quan Jianhong Zhou Angelo Genovese Vincenzo Piuri Fabio Scotti DBCG-Net: Dual Branch Calibration Guided Deep Network for UAV Images Semantic Segmentation IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing Convolutional neural network (CNN) deep learning dual-branch calibration guided network semantic segmentation unmanned aerial vehicles (UAVs) |
| title | DBCG-Net: Dual Branch Calibration Guided Deep Network for UAV Images Semantic Segmentation |
| title_full | DBCG-Net: Dual Branch Calibration Guided Deep Network for UAV Images Semantic Segmentation |
| title_fullStr | DBCG-Net: Dual Branch Calibration Guided Deep Network for UAV Images Semantic Segmentation |
| title_full_unstemmed | DBCG-Net: Dual Branch Calibration Guided Deep Network for UAV Images Semantic Segmentation |
| title_short | DBCG-Net: Dual Branch Calibration Guided Deep Network for UAV Images Semantic Segmentation |
| title_sort | dbcg net dual branch calibration guided deep network for uav images semantic segmentation |
| topic | Convolutional neural network (CNN) deep learning dual-branch calibration guided network semantic segmentation unmanned aerial vehicles (UAVs) |
| url | https://ieeexplore.ieee.org/document/10475391/ |
| work_keys_str_mv | AT chaoyunmai dbcgnetdualbranchcalibrationguideddeepnetworkforuavimagessemanticsegmentation AT yibowu dbcgnetdualbranchcalibrationguideddeepnetworkforuavimagessemanticsegmentation AT yikuizhai dbcgnetdualbranchcalibrationguideddeepnetworkforuavimagessemanticsegmentation AT haoquan dbcgnetdualbranchcalibrationguideddeepnetworkforuavimagessemanticsegmentation AT jianhongzhou dbcgnetdualbranchcalibrationguideddeepnetworkforuavimagessemanticsegmentation AT angelogenovese dbcgnetdualbranchcalibrationguideddeepnetworkforuavimagessemanticsegmentation AT vincenzopiuri dbcgnetdualbranchcalibrationguideddeepnetworkforuavimagessemanticsegmentation AT fabioscotti dbcgnetdualbranchcalibrationguideddeepnetworkforuavimagessemanticsegmentation |