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

In this paper, we introduce 3D-CSL, a compact pipeline for Near-DuplicateVideo Retrieval (NDVR), and explore a novel self-supervised learning strategyfor video similarity learning. Most previous methods only extract video spatialfeatures from frames separately and then design kinds of complex mechanisms tolearn the temporal correlations among frame features. However, parts ofspatiotemporal dependencies have already been lost. To address this, our 3D-CSLextracts global spatiotemporal dependencies in videos end-to-end with a 3Dtransformer and find a good balance between efficiency and effectiveness bymatching on clip-level. Furthermore, we propose a two-stage self-supervisedsimilarity learning strategy to optimize the entire network. Firstly, wepropose PredMAE to pretrain the 3D transformer with video prediction task;Secondly, ShotMix, a novel video-specific augmentation, and FCS loss, a noveltriplet loss, are proposed further promote the similarity learning results. Theexperiments on FIVR-200K and CC_WEB_VIDEO demonstrate the superiority andreliability of our method, which achieves the state-of-the-art performance onclip-level NDVR.

3D-CSL: self-supervised 3D context similarity learning for Near-Duplicate Video Retrieval