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

Recent years have seen an increased interest in establishing associationbetween faces and voices of celebrities leveraging audio-visual informationfrom YouTube. Prior works adopt metric learning methods to learn an embeddingspace that is amenable for associated matching and verification tasks. Albeitshowing some progress, such formulations are, however, restrictive due todependency on distance-dependent margin parameter, poor run-time trainingcomplexity, and reliance on carefully crafted negative mining procedures. Inthis work, we hypothesize that an enriched representation coupled with aneffective yet efficient supervision is important towards realizing adiscriminative joint embedding space for face-voice association tasks. To thisend, we propose a light-weight, plug-and-play mechanism that exploits thecomplementary cues in both modalities to form enriched fused embeddings andclusters them based on their identity labels via orthogonality constraints. Wecoin our proposed mechanism as fusion and orthogonal projection (FOP) andinstantiate in a two-stream network. The overall resulting framework isevaluated on VoxCeleb1 and MAV-Celeb datasets with a multitude of tasks,including cross-modal verification and matching. Results reveal that our methodperforms favourably against the current state-of-the-art methods and ourproposed formulation of supervision is more effective and efficient than theones employed by the contemporary methods. In addition, we leverage cross-modalverification and matching tasks to analyze the impact of multiple languages onface-voice association. Code is available:\url{https://github.com/msaadsaeed/FOP}

Learning Branched Fusion and Orthogonal Projection for Face-Voice Association