DC Field | Value | Language |
---|---|---|
dc.contributor.advisor | Yoon, Sung-Eui | - |
dc.contributor.advisor | 윤성의 | - |
dc.contributor.author | Cho, Yoonki | - |
dc.date.accessioned | 2022-04-27T19:32:05Z | - |
dc.date.available | 2022-04-27T19:32:05Z | - |
dc.date.issued | 2021 | - |
dc.identifier.uri | http://library.kaist.ac.kr/search/detail/view.do?bibCtrlNo=948465&flag=dissertation | en_US |
dc.identifier.uri | http://hdl.handle.net/10203/296139 | - |
dc.description | 학위논문(석사) - 한국과학기술원 : 전산학부, 2021.2,[iii, 17 p. :] | - |
dc.description.abstract | Unsupervised person re-identification (re-ID) aims to retrieve the same person across different camera-views for a given query without any labels. Recently, state-of-the-art methods that leverage pseudo-labels through clustering achieved competitive results. Despite improvements from the above methods, unreliable clusters and noisy pseudo-labels have remained problems. In this work, we propose Part-Guide Teaching (PGT) framework that effectively utilizes part-local features to address these problems. The proposed framework performs robust clustering based on part-wise relationships and refines hard pseudo-labels by guidance from part-local feature classifiers. Our refinement process works online in a self-teaching manner, which does not require additional teacher networks that need a high computational cost in the training stage. Experiment on Market-1501 and DukeMTMC-reID demonstrates the superiority of the proposed methods. | - |
dc.language | eng | - |
dc.publisher | 한국과학기술원 | - |
dc.subject | Person re-identification▼aImage search▼aUnsupervised learning▼aNoisy label | - |
dc.subject | 사람 재인식▼a이미지 검색▼a비지도 학습▼a노이지 레이블 | - |
dc.title | Part-guide teaching: part-guide pseudo-label self refinement for unsupervised person re-identification | - |
dc.title.alternative | 비지도 사람 재식별을 위한 부분 지도 의사 레이블 자가 정제 | - |
dc.type | Thesis(Master) | - |
dc.identifier.CNRN | 325007 | - |
dc.description.department | 한국과학기술원 :전산학부, | - |
dc.contributor.alternativeauthor | 조윤기 | - |
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.