Unsupervised Domain Adaptation for Visual Perception in Various Environments
- Title
- Unsupervised Domain Adaptation for Visual Perception in Various Environments
- Alternative Title
- 다양한 환경에서의 시각적 인식을 위한 비지도 도메인 적응 방법
- Author(s)
- Changjae Kim
- DGIST Authors
- Changjae Kim ; Sunghoon Im ; Kyong Hwan Jin
- Advisor
- 임성훈
- Co-Advisor(s)
- Kyong Hwan Jin
- Issued Date
- 2023
- Awarded Date
- 2023-02-01
- Type
- Thesis
- Description
- Multi-Target Domain Adaptation,Image-to-Image Translation,Semantic Segmentation,Cross-Domain Correspondence Matching,Cross-Domain Feature Consistency
- Table Of Contents
-
Ⅰ. Introduction 1
Ⅱ. Related Work 3
Ⅲ. Proposed Method 5
3.1 Class-Wise Image Translation 6
3.1.1 High-Precision Pseudo labeling 7
3.1.2 Attribute Transfer 8
3.2 Cross-Domain Feature Consistency for Domain Alignment 9
Ⅳ. Experiments 10
4.1 Datasets 10
4.2 Implementation Details 11
4.3 Synthetic-to-Real Adaptation 11
4.4 Real-to-Real Adaptation 18
4.5 Ablation study on Cross-Domain Feature Consistency 20
Ⅴ. Conclusion 21
Ⅵ. References 22
Ⅶ. 요약문 27
- URI
-
http://hdl.handle.net/20.500.11750/45751
http://dgist.dcollection.net/common/orgView/200000653655
- Degree
- Master
- Publisher
- DGIST
- Related Researcher
-
-
Im, Sunghoon
- Research Interests Computer Vision; Deep Learning; Robot Vision
-
- Files in This Item:
-
There are no files associated with this item.
- Appears in Collections:
- Department of Electrical Engineering and Computer Science Theses Master
Items in Repository are protected by copyright, with all rights reserved, unless otherwise indicated.