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

Self-supervised methods have shown remarkable progress in learning high-levelsemantics and low-level temporal correspondence. Building on these results, wetake one step further and explore the possibility of integrating these twofeatures to enhance object-centric representations. Our preliminary experimentsindicate that query slot attention can extract different semantic componentsfrom the RGB feature map, while random sampling based slot attention canexploit temporal correspondence cues between frames to assist instanceidentification. Motivated by this, we propose a novel semantic-aware maskedslot attention on top of the fused semantic features and correspondence maps.It comprises two slot attention stages with a set of shared learnable Gaussiandistributions. In the first stage, we use the mean vectors as slotinitialization to decompose potential semantics and generate semanticsegmentation masks through iterative attention. In the second stage, for eachsemantics, we randomly sample slots from the corresponding Gaussiandistribution and perform masked feature aggregation within the semantic area toexploit temporal correspondence patterns for instance identification. We adoptsemantic- and instance-level temporal consistency as self-supervision toencourage temporally coherent object-centric representations. Our modeleffectively identifies multiple object instances with semantic structure,reaching promising results on unsupervised video object discovery. Furthermore,we achieve state-of-the-art performance on dense label propagation tasks,demonstrating the potential for object-centric analysis. The code is releasedat https://github.com/shvdiwnkozbw/SMTC.

Semantics Meets Temporal Correspondence: Self-supervised Object-centric Learning in Videos