Multitask Adaptation for Unlabeled Domain Using Multiple Single-Task Domains
Semantic segmentation and depth estimation tasks are crucial for autonomous driving systems, but obtaining their labels from real-world datasets is costly. To address the problem, we developed a multitask domain adaptation that uses various labeled datasets with distinct tasks to adapt the multitask...
Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2024-01-01
|
| Series: | IEEE Access |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/10804160/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1846102169308102656 |
|---|---|
| author | Youngwook Kang Hawook Jeong Junsup Shin Jongwon Choi |
| author_facet | Youngwook Kang Hawook Jeong Junsup Shin Jongwon Choi |
| author_sort | Youngwook Kang |
| collection | DOAJ |
| description | Semantic segmentation and depth estimation tasks are crucial for autonomous driving systems, but obtaining their labels from real-world datasets is costly. To address the problem, we developed a multitask domain adaptation that uses various labeled datasets with distinct tasks to adapt the multitask model for the unlabeled domain. The proposed framework can handle multiple source domains containing various task labels, which allows us to extend the combinations of acceptable source datasets in contrast to the previous multitask domain adaptation methods. We suggest using the ‘TripleMix’ approach to obtain the integrated features from the three separate domains, including two labeled domains and one unlabeled domain. In addition, we design a task correlation network that trains multiple tasks through attentional correlation, increasing the synergies between various tasks. To validate the proposed algorithm’s state-of-the-art performance based on the interactions of the different domains and tasks, we analyze it using a variety of dataset combinations that consider two virtual domains and one real-world target domain. |
| format | Article |
| id | doaj-art-e3d03cf28fa34e8db0f7c29db43450a1 |
| institution | Kabale University |
| issn | 2169-3536 |
| language | English |
| publishDate | 2024-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Access |
| spelling | doaj-art-e3d03cf28fa34e8db0f7c29db43450a12024-12-28T00:01:26ZengIEEEIEEE Access2169-35362024-01-011219464619465610.1109/ACCESS.2024.351860010804160Multitask Adaptation for Unlabeled Domain Using Multiple Single-Task DomainsYoungwook Kang0https://orcid.org/0000-0003-1544-0888Hawook Jeong1https://orcid.org/0009-0003-5935-7055Junsup Shin2https://orcid.org/0000-0003-3280-1622Jongwon Choi3https://orcid.org/0000-0001-9753-8760Department of Advanced Imaging, Chung-Ang University, Seoul, South KoreaRideFlux Inc., Jeju, South KoreaDepartment of Advanced Imaging, Chung-Ang University, Seoul, South KoreaDepartment of Advanced Imaging, Chung-Ang University, Seoul, South KoreaSemantic segmentation and depth estimation tasks are crucial for autonomous driving systems, but obtaining their labels from real-world datasets is costly. To address the problem, we developed a multitask domain adaptation that uses various labeled datasets with distinct tasks to adapt the multitask model for the unlabeled domain. The proposed framework can handle multiple source domains containing various task labels, which allows us to extend the combinations of acceptable source datasets in contrast to the previous multitask domain adaptation methods. We suggest using the ‘TripleMix’ approach to obtain the integrated features from the three separate domains, including two labeled domains and one unlabeled domain. In addition, we design a task correlation network that trains multiple tasks through attentional correlation, increasing the synergies between various tasks. To validate the proposed algorithm’s state-of-the-art performance based on the interactions of the different domains and tasks, we analyze it using a variety of dataset combinations that consider two virtual domains and one real-world target domain.https://ieeexplore.ieee.org/document/10804160/Domain adaptationdepth estimationmultitask learningsemantic segmentation |
| spellingShingle | Youngwook Kang Hawook Jeong Junsup Shin Jongwon Choi Multitask Adaptation for Unlabeled Domain Using Multiple Single-Task Domains IEEE Access Domain adaptation depth estimation multitask learning semantic segmentation |
| title | Multitask Adaptation for Unlabeled Domain Using Multiple Single-Task Domains |
| title_full | Multitask Adaptation for Unlabeled Domain Using Multiple Single-Task Domains |
| title_fullStr | Multitask Adaptation for Unlabeled Domain Using Multiple Single-Task Domains |
| title_full_unstemmed | Multitask Adaptation for Unlabeled Domain Using Multiple Single-Task Domains |
| title_short | Multitask Adaptation for Unlabeled Domain Using Multiple Single-Task Domains |
| title_sort | multitask adaptation for unlabeled domain using multiple single task domains |
| topic | Domain adaptation depth estimation multitask learning semantic segmentation |
| url | https://ieeexplore.ieee.org/document/10804160/ |
| work_keys_str_mv | AT youngwookkang multitaskadaptationforunlabeleddomainusingmultiplesingletaskdomains AT hawookjeong multitaskadaptationforunlabeleddomainusingmultiplesingletaskdomains AT junsupshin multitaskadaptationforunlabeleddomainusingmultiplesingletaskdomains AT jongwonchoi multitaskadaptationforunlabeleddomainusingmultiplesingletaskdomains |