DSpace Repository

Instance-level Heterogeneous Domain Adaptation for Limited-labeled Sketch-to-Photo Retrieval

Show simple item record

dc.contributor.author Yang, Fan
dc.contributor.author Wu, Yang
dc.contributor.author Wang, Zheng
dc.contributor.author Li, Xiang
dc.contributor.author Sakti, Sakriani
dc.contributor.author Nakamura, Satoshi
dc.date.accessioned 2020-11-26T07:38:43Z
dc.date.available 2020-11-26T07:38:43Z
dc.date.issued 2020-07-15
dc.identifier.uri http://hdl.handle.net/10061/14160
dc.description.abstract Although sketch-to-photo retrieval has a wide range of applications, it is costly to obtain paired and rich-labeled ground truth. Differently, photo retrieval data is easier to acquire. Therefore, previous works pre-train their models on rich-labeled photo retrieval data (i.e., source domain) and then fine-tune them on the limited-labeled sketch-to-photo retrieval data (i.e., target domain). However, without co-training source and target data, source domain knowledge might be forgotten during the fine-tuning process, while simply co-training them may cause negative transfer due to domain gaps. Moreover, identity label spaces of source data and target data are generally disjoint and therefore conventional category-level Domain Adaptation (DA) is not directly applicable. To address these issues, we propose an Instance-level Heterogeneous Domain Adaptation (IHDA) framework. We apply the fine-tuning strategy for identity label learning, aiming to transfer the instance-level knowledge in an inductive transfer manner. Meanwhile, labeled attributes from the source data are selected to form a shared label space for source and target domains. Guided by shared attributes, DA is utilized to bridge cross-data domain gaps and heterogeneous domain gaps, which transfers instance-level knowledge in a transductive transfer manner. Experiments show that our method has set a new state of the art on three sketch-to-photo image retrieval benchmarks without extra annotations, which opens the door to train more effective models on limited-labeled heterogeneous image retrieval tasks. Our method is implemented by Pytorch framework and our code will be released. ja_JP
dc.language.iso en ja_JP
dc.publisher IEEE ja_JP
dc.rights © 2020IEEE ja_JP
dc.rights 出版社許諾条件により、本文は2021年7月11日以降に公開 ja
dc.rights 出版社許諾条件により、本文は2021年7月16日以降に公開 ja
dc.subject Task analysis ja_JP
dc.subject Training data ja_JP
dc.subject Data models ja_JP
dc.subject Image retrieval ja_JP
dc.subject Connectors ja_JP
dc.subject Training ja_JP
dc.subject Entropy ja_JP
dc.title Instance-level Heterogeneous Domain Adaptation for Limited-labeled Sketch-to-Photo Retrieval ja_JP
dc.type.nii Journal Article ja_JP
dc.contributor.transcription ナカムラ, サトシ
dc.contributor.alternative 中村, 哲
dc.textversion author ja_JP
dc.identifier.eissn 1941-0077
dc.identifier.jtitle IEEE Transactions on Multimedia ja_JP
dc.relation.doi 10.1109/TMM.2020.3009476 ja_JP
dc.identifier.NAIST-ID 85620342 ja_JP
dc.identifier.NAIST-ID 74651555 ja_JP
dc.identifier.NAIST-ID 73297715 ja_JP
dc.identifier.NAIST-ID 73296626 ja_JP


Files in this item

This item appears in the following Collection(s)

Show simple item record

Search DSpace


Advanced Search

Browse

My Account