Show simple item record

dc.contributor.author
Nitsch, Julia
dc.contributor.author
Nieto, Juan
dc.contributor.author
Siegwart, Roland
dc.contributor.author
Schmidt, Max
dc.contributor.author
Cadena, Cesar
dc.date.accessioned
2021-07-21T12:07:00Z
dc.date.available
2021-07-15T10:16:56Z
dc.date.available
2021-07-21T12:07:00Z
dc.date.issued
2020
dc.identifier.isbn
978-1-7281-6673-5
en_US
dc.identifier.isbn
978-1-7281-6672-8
en_US
dc.identifier.isbn
978-1-7281-6674-2
en_US
dc.identifier.other
10.1109/IV47402.2020.9304669
en_US
dc.identifier.uri
http://hdl.handle.net/20.500.11850/494763
dc.description.abstract
LiDAR sensors are crucial in automotive perception for accurate object detection. However, LiDAR data is hard to interpret for humans and consequently time-consuming to label. Whereas camera data is easy interpretable and thus, comparably simpler to label. Within this work we present a transductive transfer learning approach to transfer the knowledge for the object detection task from images to point cloud data. We propose a multi-modal adversarial Auto Encoder architecture which disentangles uni-modal features into two groups: common (transferable) features, and complementary (modality-specific) features. This disentanglement is based on the hypothesis that a set of common features exist. An important point of our framework is that the disentanglement is learned in an unsupervised manner. Furthermore, the results show that only a small amount of multi-modal data is needed to learn the disentanglement, and thus to transfer the knowledge between modalities. As a result we our experiments show that training with 75% less data of the KITTI objects, the classification accuracy achieved is of 71.75%, only 3.12% less than when using the full data set. The implications of these findings can have great impact in perception pipelines based on LIDAR data.
en_US
dc.language.iso
en
en_US
dc.publisher
IEEE
en_US
dc.title
Learning Common and Transferable Feature Representations for Multi-Modal Data
en_US
dc.type
Conference Paper
dc.date.published
2021-01-08
ethz.book.title
2020 IEEE Intelligent Vehicles Symposium (IV)
en_US
ethz.pages.start
1595
en_US
ethz.pages.end
1601
en_US
ethz.event
31st IEEE Intelligent Vehicles Symposium (IV 2020) (virtual)
en_US
ethz.event.location
Las Vegas, NV, USA
en_US
ethz.event.date
October 19 - November 13, 2020
en_US
ethz.notes
Due to the Coronavirus (COVID-19) the conference was conducted virtually.
en_US
ethz.identifier.wos
ethz.publication.place
Piscataway, NJ
en_US
ethz.publication.status
published
en_US
ethz.date.deposited
2021-07-15T10:18:15Z
ethz.source
WOS
ethz.eth
yes
en_US
ethz.availability
Metadata only
en_US
ethz.rosetta.installDate
2021-07-21T12:07:06Z
ethz.rosetta.lastUpdated
2021-07-21T12:07:06Z
ethz.rosetta.versionExported
true
ethz.COinS
ctx_ver=Z39.88-2004&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.atitle=Learning%20Common%20and%20Transferable%20Feature%20Representations%20for%20Multi-Modal%20Data&rft.date=2020&rft.spage=1595&rft.epage=1601&rft.au=Nitsch,%20Julia&Nieto,%20Juan&Siegwart,%20Roland&Schmidt,%20Max&Cadena,%20Cesar&rft.isbn=978-1-7281-6673-5&978-1-7281-6672-8&978-1-7281-6674-2&rft.genre=proceeding&rft_id=info:doi/10.1109/IV47402.2020.9304669&rft.btitle=2020%20IEEE%20Intelligent%20Vehicles%20Symposium%20(IV)
 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