Semi-supervised semantic labeling of remote sensing images with improved image-level selection retraining
In recent years, image semantic segmentation technology has developed rapidly, but image annotation usually requires a significant amount of human and financial resources, especially for remote sensing image annotation, which can be expensive and sometimes even unaffordable. To address this issue, t...
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2024-05-01
|
| Series: | Alexandria Engineering Journal |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S1110016824002709 |
| _version_ | 1797203070266900480 |
|---|---|
| author | Qiongqiong Hu Yuechao Wu Ying Li |
| author_facet | Qiongqiong Hu Yuechao Wu Ying Li |
| author_sort | Qiongqiong Hu |
| collection | DOAJ |
| description | In recent years, image semantic segmentation technology has developed rapidly, but image annotation usually requires a significant amount of human and financial resources, especially for remote sensing image annotation, which can be expensive and sometimes even unaffordable. To address this issue, this paper integrates the idea of curriculum learning into the self-training method and screens reliable pseudo-labels through computing image-level confidence, significantly reducing the confirmation error problem. Furthermore, the semi-supervised model in this paper combines implicit semantic enhancement with strong data augmentation, which can reduce the coupling between the teacher model and the student model’s prediction distribution and enhance the model’s robustness. Finally, the proposed semi-supervised method is experimentally verified using the ISPRS competition dataset and compared with existing state-of-the-art (SOTA) methods. Experimental results show that the proposed semi-supervised segmentation method achieves higher segmentation accuracy compared to self-training methods. Moreover, despite not using iterative training to simplify the training process, the proposed method still yields satisfactory segmentation results. |
| first_indexed | 2024-04-24T08:13:29Z |
| format | Article |
| id | doaj.art-0233f965be344769afda65ba0e02a51b |
| institution | Directory Open Access Journal |
| issn | 1110-0168 |
| language | English |
| last_indexed | 2024-04-24T08:13:29Z |
| publishDate | 2024-05-01 |
| publisher | Elsevier |
| record_format | Article |
| series | Alexandria Engineering Journal |
| spelling | doaj.art-0233f965be344769afda65ba0e02a51b2024-04-17T04:48:39ZengElsevierAlexandria Engineering Journal1110-01682024-05-0194235247Semi-supervised semantic labeling of remote sensing images with improved image-level selection retrainingQiongqiong Hu0Yuechao Wu1Ying Li2School of Computer Science, Northwestern Polytechnical University, Xi’an, Shaanxi 710129, ChinaDepartment of Computer Technology and Application, Qinghai University, Xi’ning, Qinghai 810016, ChinaSchool of Computer Science, Northwestern Polytechnical University, Xi’an, Shaanxi 710129, China; Corresponding author.In recent years, image semantic segmentation technology has developed rapidly, but image annotation usually requires a significant amount of human and financial resources, especially for remote sensing image annotation, which can be expensive and sometimes even unaffordable. To address this issue, this paper integrates the idea of curriculum learning into the self-training method and screens reliable pseudo-labels through computing image-level confidence, significantly reducing the confirmation error problem. Furthermore, the semi-supervised model in this paper combines implicit semantic enhancement with strong data augmentation, which can reduce the coupling between the teacher model and the student model’s prediction distribution and enhance the model’s robustness. Finally, the proposed semi-supervised method is experimentally verified using the ISPRS competition dataset and compared with existing state-of-the-art (SOTA) methods. Experimental results show that the proposed semi-supervised segmentation method achieves higher segmentation accuracy compared to self-training methods. Moreover, despite not using iterative training to simplify the training process, the proposed method still yields satisfactory segmentation results.http://www.sciencedirect.com/science/article/pii/S1110016824002709Deep convolutional neural networksRemote sensing imagesSemantic labelingSemi-supervised learning |
| spellingShingle | Qiongqiong Hu Yuechao Wu Ying Li Semi-supervised semantic labeling of remote sensing images with improved image-level selection retraining Alexandria Engineering Journal Deep convolutional neural networks Remote sensing images Semantic labeling Semi-supervised learning |
| title | Semi-supervised semantic labeling of remote sensing images with improved image-level selection retraining |
| title_full | Semi-supervised semantic labeling of remote sensing images with improved image-level selection retraining |
| title_fullStr | Semi-supervised semantic labeling of remote sensing images with improved image-level selection retraining |
| title_full_unstemmed | Semi-supervised semantic labeling of remote sensing images with improved image-level selection retraining |
| title_short | Semi-supervised semantic labeling of remote sensing images with improved image-level selection retraining |
| title_sort | semi supervised semantic labeling of remote sensing images with improved image level selection retraining |
| topic | Deep convolutional neural networks Remote sensing images Semantic labeling Semi-supervised learning |
| url | http://www.sciencedirect.com/science/article/pii/S1110016824002709 |
| work_keys_str_mv | AT qiongqionghu semisupervisedsemanticlabelingofremotesensingimageswithimprovedimagelevelselectionretraining AT yuechaowu semisupervisedsemanticlabelingofremotesensingimageswithimprovedimagelevelselectionretraining AT yingli semisupervisedsemanticlabelingofremotesensingimageswithimprovedimagelevelselectionretraining |