DLR-Logo -> http://www.dlr.de
DLR Portal Home | Imprint | Privacy Policy | Contact | Deutsch
Fontsize: [-] Text [+]

Sympathy for the Details: Dense Trajectories and Hybrid Classification Architectures for Action Recognition

de Souza, César Roberto and Gaidon, Adrien and Vig, Eleonora and López, Antonio Manuel (2016) Sympathy for the Details: Dense Trajectories and Hybrid Classification Architectures for Action Recognition. In: Proceedings of the 14th European Conference on Computer Vision (ECCV), 9911 (P VII), pp. 697-716. Springer International Publishing. 14th European Conference on Computer Vision (ECCV), 11-14 October 2016, Amsterdam, NL. DOI: 10.​1007/​978-3-319-46478-7_​43 ISBN 978-3-319-46477-0 (P) 978-3-319-46478-7 (E) ISSN 0302-9743

[img] PDF

Official URL: http://link.springer.com/chapter/10.1007%2F978-3-319-46478-7_43


Action recognition in videos is a challenging task due to the complexity of the spatio-temporal patterns to model and the difficulty to acquire and learn on large quantities of video data. Deep learning, although a breakthrough for Image classification and showing promise for videos, has still not clearly superseded action recognition methods using hand-crafted features, even when training on massive datasets. In this paper, we introduce hybrid video classification architectures based on carefully designed unsupervised representations of hand-crafted spatio-temporal features classified by supervised deep networks. As we show in our experiments on five popular benchmarks for action recognition, our hybrid model combines the best of both worlds: it is data efficient (trained on 150 to 10000 short clips) and yet improves significantly on the state of the art, including recent deep models trained on millions of manually labelled images and videos.

Item URL in elib:https://elib.dlr.de/107744/
Document Type:Conference or Workshop Item (Poster)
Title:Sympathy for the Details: Dense Trajectories and Hybrid Classification Architectures for Action Recognition
AuthorsInstitution or Email of AuthorsAuthors ORCID iD
de Souza, César Robertocesar.desouza (at) xrce.xerox.comUNSPECIFIED
Gaidon, Adrienadrien.gaidon (at) xrce.xerox.comUNSPECIFIED
Vig, Eleonoraeleonora.vig (at) dlr.deUNSPECIFIED
López, Antonio Manuelantonio (at) cvc.uab.esUNSPECIFIED
Journal or Publication Title:Proceedings of the 14th European Conference on Computer Vision (ECCV)
Refereed publication:Yes
Open Access:Yes
Gold Open Access:No
In ISI Web of Science:No
DOI :10.​1007/​978-3-319-46478-7_​43
Page Range:pp. 697-716
Publisher:Springer International Publishing
Series Name:Series Lecture Notes in Computer Science
ISBN:978-3-319-46477-0 (P) 978-3-319-46478-7 (E)
Keywords:Action Recognition
Event Title:14th European Conference on Computer Vision (ECCV)
Event Location:Amsterdam, NL
Event Type:international Conference
Event Dates:11-14 October 2016
Organizer:University of Amsterdam
HGF - Research field:Aeronautics, Space and Transport
HGF - Program:Transport
HGF - Program Themes:Traffic Management (old)
DLR - Research area:Transport
DLR - Program:V VM - Verkehrsmanagement
DLR - Research theme (Project):V - Vabene++ (old)
Location: Oberpfaffenhofen
Institutes and Institutions:Remote Sensing Technology Institute > Photogrammetry and Image Analysis
Deposited On:30 Nov 2016 17:46
Last Modified:31 Jul 2019 20:04

Repository Staff Only: item control page

Help & Contact
electronic library is running on EPrints 3.3.12
Copyright © 2008-2017 German Aerospace Center (DLR). All rights reserved.