File Download
There are no files associated with this item.
Links for fulltext
(May Require Subscription)
- Publisher Website: 10.1007/978-3-031-19827-4_22
- Scopus: eid_2-s2.0-85142741130
- WOS: WOS:000903572500022
- Find via
Supplementary
- Citations:
- Appears in Collections:
Conference Paper: DecoupleNet: Decoupled Network for Domain Adaptive Semantic Segmentation
Title | DecoupleNet: Decoupled Network for Domain Adaptive Semantic Segmentation |
---|---|
Authors | |
Keywords | Semantic segmentation Unsupervised domain adaptation |
Issue Date | 23-Oct-2022 |
Publisher | Springer |
Abstract | Unsupervised domain adaptation in semantic segmentation alleviates the reliance on expensive pixel-wise annotation. It uses a labeled source domain dataset as well as unlabeled target domain images to learn a segmentation network. In this paper, we observe two main issues of existing domain-invariant learning framework. (1) Being distracted by the feature distribution alignment, the network cannot focus on the segmentation task. (2) Fitting source domain data well would compromise the target domain performance. To address these issues, we propose DecoupleNet to alleviate source domain overfitting and let the final model focus more on the segmentation task. Also, we put forward SelfDiscrimination (SD) and introduce an auxiliary classifier to learn more discriminative target domain features with pseudo labels. Finally, we propose Online Enhanced Self-Training (OEST) to contextually enhance the quality of pseudo labels in an online manner. Experiments show our method outperforms existing state-of-the-art methods. Extensive ablation studies verify the effectiveness of each component. Code is available at https://github.com/dvlab- research/DecoupleNet. |
Persistent Identifier | http://hdl.handle.net/10722/333854 |
ISSN | 2023 SCImago Journal Rankings: 0.606 |
ISI Accession Number ID |
DC Field | Value | Language |
---|---|---|
dc.contributor.author | Lai, X | - |
dc.contributor.author | Tian, ZT | - |
dc.contributor.author | Xu, XG | - |
dc.contributor.author | Chen, YC | - |
dc.contributor.author | Liu, S | - |
dc.contributor.author | Zhao, HS | - |
dc.contributor.author | Wang, LW | - |
dc.contributor.author | Jia, JY | - |
dc.date.accessioned | 2023-10-06T08:39:37Z | - |
dc.date.available | 2023-10-06T08:39:37Z | - |
dc.date.issued | 2022-10-23 | - |
dc.identifier.issn | 0302-9743 | - |
dc.identifier.uri | http://hdl.handle.net/10722/333854 | - |
dc.description.abstract | Unsupervised domain adaptation in semantic segmentation alleviates the reliance on expensive pixel-wise annotation. It uses a labeled source domain dataset as well as unlabeled target domain images to learn a segmentation network. In this paper, we observe two main issues of existing domain-invariant learning framework. (1) Being distracted by the feature distribution alignment, the network cannot focus on the segmentation task. (2) Fitting source domain data well would compromise the target domain performance. To address these issues, we propose DecoupleNet to alleviate source domain overfitting and let the final model focus more on the segmentation task. Also, we put forward SelfDiscrimination (SD) and introduce an auxiliary classifier to learn more discriminative target domain features with pseudo labels. Finally, we propose Online Enhanced Self-Training (OEST) to contextually enhance the quality of pseudo labels in an online manner. Experiments show our method outperforms existing state-of-the-art methods. Extensive ablation studies verify the effectiveness of each component. Code is available at https://github.com/dvlab- research/DecoupleNet. | - |
dc.language | eng | - |
dc.publisher | Springer | - |
dc.relation.ispartof | 17th European Conference on Computer Vision (ECCV) (23/10/2022, Tel Aviv) | - |
dc.subject | Semantic segmentation | - |
dc.subject | Unsupervised domain adaptation | - |
dc.title | DecoupleNet: Decoupled Network for Domain Adaptive Semantic Segmentation | - |
dc.type | Conference_Paper | - |
dc.identifier.doi | 10.1007/978-3-031-19827-4_22 | - |
dc.identifier.scopus | eid_2-s2.0-85142741130 | - |
dc.identifier.volume | 13693 | - |
dc.identifier.spage | 369 | - |
dc.identifier.epage | 387 | - |
dc.identifier.eissn | 1611-3349 | - |
dc.identifier.isi | WOS:000903572500022 | - |
dc.publisher.place | CHAM | - |
dc.identifier.issnl | 0302-9743 | - |