A Context-Aware Loss Function for Action Spotting in Soccer Videos

Anthony Cioppa, Adrien Deliege, Silvio Giancola, Bernard Ghanem, Marc Van Droogenbroeck, Rikke Gade, Thomas B. Moeslund

Publikation: Bidrag til bog/antologi/rapport/konference proceedingKonferenceartikel i proceedingForskningpeer review

1 Citationer (Scopus)
10 Downloads (Pure)

Abstrakt

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.

OriginalsprogEngelsk
Titel2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)
Antal sider11
ForlagIEEE
Publikationsdatojun. 2020
Sider13123-13133
ISBN (Trykt)978-1-7281-7169-2
ISBN (Elektronisk)978-1-7281-7168-5
DOI
StatusUdgivet - jun. 2020
Begivenhed2020 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) - Seattle, USA
Varighed: 14 jun. 202019 jun. 2020

Konference

Konference2020 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)
LandUSA
BySeattle
Periode14/06/202019/06/2020
NavnI E E E Conference on Computer Vision and Pattern Recognition. Proceedings
ISSN1063-6919

Fingeraftryk Dyk ned i forskningsemnerne om 'A Context-Aware Loss Function for Action Spotting in Soccer Videos'. Sammen danner de et unikt fingeraftryk.

Citationsformater