Show simple item record

dc.contributor.author
Dall'Asen, Nicola
dc.contributor.author
Wang, Yiming
dc.contributor.author
Tang, Hao
dc.contributor.author
Zanella, Luca
dc.contributor.author
Ricci, Elisa
dc.contributor.editor
Sclaroff, Stan
dc.contributor.editor
Distante, Cosimo
dc.contributor.editor
Leo, Marco
dc.contributor.editor
Farinella, Giovanni M.
dc.contributor.editor
Tombari, Federico
dc.date.accessioned
2022-11-09T14:30:39Z
dc.date.available
2022-11-05T04:05:56Z
dc.date.available
2022-11-09T14:30:39Z
dc.date.issued
2022
dc.identifier.isbn
978-3-031-06430-2
en_US
dc.identifier.isbn
978-3-031-06429-6
en_US
dc.identifier.issn
0302-9743
dc.identifier.issn
1611-3349
dc.identifier.other
10.1007/978-3-031-06430-2_42
en_US
dc.identifier.uri
http://hdl.handle.net/20.500.11850/579607
dc.description.abstract
We propose AnonyGAN, a GAN-based solution for face anonymisation which replaces the visual information corresponding to a source identity with a condition identity provided as any single image. With the goal to maintain the geometric attributes of the source face, i.e., the facial pose and expression, and to promote more natural face generation, we propose to exploit a Bipartite Graph to explicitly model the relations between the facial landmarks of the source identity and the ones of the condition identity through a deep model. We further propose a landmark attention model to relax the manual selection of facial landmarks, allowing the network to weight the landmarks for the best visual naturalness and pose preservation. Finally, to facilitate the appearance learning, we propose a hybrid training strategy to address the challenge caused by the lack of direct pixel-level supervision. We evaluate our method and its variants on two public datasets, CelebA and LFW, in terms of visual naturalness, facial pose preservation and of its impacts on face detection and re-identification. We prove that AnonyGAN significantly outperforms the state-of-the-art methods in terms of visual naturalness, face detection and pose preservation. Code and pretrained model are available at https://github.com/Fodark/anonygan.
en_US
dc.language.iso
en
en_US
dc.publisher
Springer
en_US
dc.title
Graph-Based Generative Face Anonymisation with Pose Preservation
en_US
dc.type
Conference Paper
dc.date.published
2022-05-17
ethz.book.title
Image Analysis and Processing – ICIAP 2022
en_US
ethz.journal.title
Lecture Notes in Computer Science
ethz.journal.volume
13232
en_US
ethz.journal.abbreviated
LNCS
ethz.pages.start
503
en_US
ethz.pages.end
515
en_US
ethz.event
21st International Conference on Image Analysis and Processing (ICIAP 2022)
en_US
ethz.event.location
Lecce, Italy
en_US
ethz.event.date
May 23-27, 2022
en_US
ethz.identifier.wos
ethz.publication.place
Cham
en_US
ethz.publication.status
published
en_US
ethz.date.deposited
2022-11-05T04:06:01Z
ethz.source
WOS
ethz.eth
yes
en_US
ethz.availability
Metadata only
en_US
ethz.rosetta.installDate
2022-11-09T14:30:40Z
ethz.rosetta.lastUpdated
2022-11-09T14:30:40Z
ethz.rosetta.versionExported
true
ethz.COinS
ctx_ver=Z39.88-2004&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.atitle=Graph-Based%20Generative%20Face%20Anonymisation%20with%20Pose%20Preservation&rft.jtitle=Lecture%20Notes%20in%20Computer%20Science&rft.date=2022&rft.volume=13232&rft.spage=503&rft.epage=515&rft.issn=0302-9743&1611-3349&rft.au=Dall'Asen,%20Nicola&Wang,%20Yiming&Tang,%20Hao&Zanella,%20Luca&Ricci,%20Elisa&rft.isbn=978-3-031-06430-2&978-3-031-06429-6&rft.genre=proceeding&rft_id=info:doi/10.1007/978-3-031-06430-2_42&rft.btitle=Image%20Analysis%20and%20Processing%20%E2%80%93%20ICIAP%202022
 Search print copy at ETH Library

Files in this item

FilesSizeFormatOpen in viewer

There are no files associated with this item.

Publication type

Show simple item record