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

Audio-Visual Emotion Recognition (AVER) has garnered increasing attention inrecent years for its critical role in creating emotion-ware intelligentmachines. Previous efforts in this area are dominated by the supervisedlearning paradigm. Despite significant progress, supervised learning is meetingits bottleneck due to the longstanding data scarcity issue in AVER. Motivatedby recent advances in self-supervised learning, we propose HierarchicalContrastive Masked Autoencoder (HiCMAE), a novel self-supervised framework thatleverages large-scale self-supervised pre-training on vast unlabeledaudio-visual data to promote the advancement of AVER. Following prior arts inself-supervised audio-visual representation learning, HiCMAE adopts two primaryforms of self-supervision for pre-training, namely masked data modeling andcontrastive learning. Unlike them which focus exclusively on top-layerrepresentations while neglecting explicit guidance of intermediate layers,HiCMAE develops a three-pronged strategy to foster hierarchical audio-visualfeature learning and improve the overall quality of learned representations. Toverify the effectiveness of HiCMAE, we conduct extensive experiments on 9datasets covering both categorical and dimensional AVER tasks. Experimentalresults show that our method significantly outperforms state-of-the-artsupervised and self-supervised audio-visual methods, which indicates thatHiCMAE is a powerful audio-visual emotion representation learner. Codes andmodels will be publicly available at https://github.com/sunlicai/HiCMAE.

HiCMAE: Hierarchical Contrastive Masked Autoencoder for Self-Supervised Audio-Visual Emotion Recognition