DC Field | Value | Language |
---|---|---|
dc.contributor.advisor | Kim, Junmo | - |
dc.contributor.advisor | 김준모 | - |
dc.contributor.author | Han, Sangeun | - |
dc.date.accessioned | 2022-04-27T19:31:04Z | - |
dc.date.available | 2022-04-27T19:31:04Z | - |
dc.date.issued | 2021 | - |
dc.identifier.uri | http://library.kaist.ac.kr/search/detail/view.do?bibCtrlNo=948993&flag=dissertation | en_US |
dc.identifier.uri | http://hdl.handle.net/10203/295957 | - |
dc.description | 학위논문(석사) - 한국과학기술원 : 전기및전자공학부, 2021.2,[iii, 15 p. :] | - |
dc.description.abstract | Existing image inpainting methods do not utilize identity information for face completion, producing images of different identities. Considering that identity preservation is important in many real-world face editing applications, we propose a task-specific approach for identity-aware face completion, which is guided by a single reference image containing identity information. Experimental results show that our approach improves the visual quality of the completion results while preserving identity. | - |
dc.language | eng | - |
dc.publisher | 한국과학기술원 | - |
dc.subject | face completion▼aimage inpainting▼agenerative adversarial network▼acomputer vision▼adeep learning | - |
dc.subject | 얼굴 인페인팅▼a영상 인페인팅▼a생성적 적대 신경망▼a컴퓨터 비전▼a심층 학습 | - |
dc.title | Identity-aware face completion for face editing applications | - |
dc.title.alternative | 얼굴 보정 어플리케이션을 위한 고유성 보존 인페인팅 | - |
dc.type | Thesis(Master) | - |
dc.identifier.CNRN | 325007 | - |
dc.description.department | 한국과학기술원 :전기및전자공학부, | - |
dc.contributor.alternativeauthor | 한상은 | - |
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.