Paper published in a journal (Scientific congresses and symposiums)
A Context-Aware Loss Function for Action Spotting in Soccer Videos
Cioppa, Anthony; Deliège, Adrien; Giancola, Silvio et al.
2020In IEEE Conference on Computer Vision and Pattern Recognition. Proceedings, p. 13123-13133
Peer reviewed
 

Files


Full Text
Cioppa2020AContextAware.pdf
Author postprint (1.6 MB)
A Context-Aware Loss Function for Action Spotting in Soccer Videos
Download
Annexes
Cioppa2020AContextAware-supp.pdf
Publisher postprint (4.22 MB)
Supplementary material
Download

All documents in ORBi are protected by a user license.

Send to



Details



Keywords :
action spotting; artificial intelligence; soccer videos; deep learning; context-aware loss; highlights generation; sports; sport; football
Abstract :
[en] In video understanding, action spotting consists in temporally localizing human-induced events annotated with single timestamps. In this paper, we propose a novel loss function that specifically considers the temporal context naturally present around each action, rather than focusing on the single annotated frame to spot. We benchmark our loss on a large dataset of soccer videos, SoccerNet, and achieve an improvement of 12.8% over the baseline. We show the generalization capability of our loss for generic activity proposals and detection on ActivityNet, by spotting the beginning and the end of each activity. Furthermore, we provide an extended ablation study and display challenging cases for action spotting in soccer videos. Finally, we qualitatively illustrate how our loss induces a precise temporal understanding of actions and show how such semantic knowledge can be used for automatic highlights generation.
Research center :
Montefiore Institute of Electrical Engineering and Computer Science - Montefiore Institute ; Telim
TELIM
Disciplines :
Computer science
Author, co-author :
Cioppa, Anthony  ;  Université de Liège - ULiège > Dép. d'électric., électron. et informat. (Inst.Montefiore) > Télécommunications
Deliège, Adrien  ;  Université de Liège - ULiège > Dép. d'électric., électron. et informat. (Inst.Montefiore) > Télécommunications
Giancola, Silvio 
Ghanem, Bernard
Van Droogenbroeck, Marc  ;  Université de Liège - ULiège > Dép. d'électric., électron. et informat. (Inst.Montefiore) > Télécommunications
Gade, Rikke
Moeslund, Thomas B.
 These authors have contributed equally to this work.
Language :
English
Title :
A Context-Aware Loss Function for Action Spotting in Soccer Videos
Publication date :
June 2020
Event name :
IEEE Conference on Computer Vision and Pattern Recognition
Event organizer :
IEEE
Event place :
Seattle, United States - Washington
Event date :
from 14-06-2020 to 19-06-2020
Audience :
International
Journal title :
IEEE Conference on Computer Vision and Pattern Recognition. Proceedings
ISSN :
1063-6919
eISSN :
2575-7075
Publisher :
IEEE Computer Society, Washington, United States - District of Columbia
Pages :
13123-13133
Peer reviewed :
Peer reviewed
Name of the research project :
DeepSport
Funders :
DGTRE - Région wallonne. Direction générale des Technologies, de la Recherche et de l'Énergie [BE]
FRIA - Fonds pour la Formation à la Recherche dans l'Industrie et dans l'Agriculture [BE]
KAUST Office of Sponsored Research
Commentary :
Accepted for publication in CVPR2020 main conference. Code source at https://github.com/cioppaanthony/context-aware-loss
Available on ORBi :
since 04 December 2019

Statistics


Number of views
223 (56 by ULiège)
Number of downloads
184 (26 by ULiège)

Scopus citations®
 
47
Scopus citations®
without self-citations
38
OpenCitations
 
36

Bibliography


Similar publications



Contact ORBi