Show simple item record

dc.contributor.authorZhao, Chen
dc.contributor.authorThabet, Ali Kassem
dc.contributor.authorGhanem, Bernard
dc.date.accessioned2020-12-02T07:52:53Z
dc.date.available2020-12-02T07:52:53Z
dc.date.issued2020-11-30
dc.identifier.urihttp://hdl.handle.net/10754/666221
dc.description.abstractTemporal action localization (TAL) in videos is a challenging task, especially due to the large scale variation of actions. In the data, short actions usually occupy the major proportion, but have the lowest performance with all current methods. In this paper, we confront the challenge of short actions and propose a multi-level cross-scale solution dubbed as video self-stitching graph network (VSGN). We have two key components in VSGN: video self-stitching (VSS) and cross-scale graph pyramid network (xGPN). In VSS, we focus on a short period of a video and magnify it along the temporal dimension to obtain a larger scale. By our self-stitching approach, we are able to utilize the original clip and its magnified counterpart in one input sequence to take advantage of the complementary properties of both scales. The xGPN component further exploits the cross-scale correlations by a pyramid of cross-scale graph networks, each containing a hybrid temporal-graph module to aggregate features from across scales as well as within the same scale. Our VSGN not only enhances the feature representations, but also generates more positive anchors for short actions and more short training samples. Experiments demonstrate that VSGN obviously improves the localization performance of short actions as well as achieving the state-of-the-art overall performance on ActivityNet-v1.3, reaching an average mAP of 35.07 %.
dc.publisherarXiv
dc.relation.urlhttps://arxiv.org/pdf/2011.14598
dc.rightsArchived with thanks to arXiv
dc.titleVideo Self-Stitching Graph Network for Temporal Action Localization
dc.typePreprint
dc.contributor.departmentVisual Computing Center (VCC)
dc.contributor.departmentComputer, Electrical and Mathematical Sciences and Engineering (CEMSE) Division
dc.contributor.departmentElectrical Engineering Program
dc.eprint.versionPre-print
dc.identifier.arxivid2011.14598
kaust.personZhao, Chen
kaust.personThabet, Ali Kassem
kaust.personGhanem, Bernard
refterms.dateFOA2020-12-02T07:54:24Z


Files in this item

Thumbnail
Name:
Preprintfile1.pdf
Size:
1.887Mb
Format:
PDF
Description:
Pre-print

This item appears in the following Collection(s)

Show simple item record