English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Temporal grounding aims to retrieve moments of the described event within anuntrimmed video by a language query. Typically, existing methods assumeannotations are precise and unique, yet one query may describe multiple momentsin many cases. Hence, simply taking it as a one-vs-one mapping task andstriving to match single-label annotations will inevitably introduce falsenegatives during optimization. In this study, we reformulate this task as aone-vs-many optimization problem under the condition of single positive labels.The unlabeled moments are considered unobserved rather than negative, and weexplore mining potential positive moments to assist in multiple momentretrieval. In this setting, we propose a novel Diverse Temporal Groundingframework, termed DTG-SPL, which mainly consists of a positive momentestimation (PME) module and a diverse moment regression (DMR) module. PMEleverages semantic reconstruction information and an expected positiveregularization to uncover potential positive moments in an online fashion.Under the supervision of these pseudo positives, DMR is able to localizediverse moments in parallel that meet different users. The entire frameworkallows for end-to-end optimization as well as fast inference. Extensiveexperiments on Charades-STA and ActivityNet Captions show that our methodachieves superior performance in terms of both single-label and multi-labelmetrics.

Towards Diverse Temporal Grounding under Single Positive Labels