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 the latest social networks, more and more people prefer to express theiremotions in videos through text, speech, and rich facial expressions.Multimodal video emotion analysis techniques can help understand users' innerworld automatically based on human expressions and gestures in images, tones invoices, and recognized natural language. However, in the existing research, theacoustic modality has long been in a marginal position as compared to visualand textual modalities. That is, it tends to be more difficult to improve thecontribution of the acoustic modality for the whole multimodal emotionrecognition task. Besides, although better performance can be obtained byintroducing common deep learning methods, the complex structures of thesetraining models always result in low inference efficiency, especially whenexposed to high-resolution and long-length videos. Moreover, the lack of afully end-to-end multimodal video emotion recognition system hinders itsapplication. In this paper, we designed a fully multimodal video-to-emotionsystem (named FV2ES) for fast yet effective recognition inference, whosebenefits are threefold: (1) The adoption of the hierarchical attention methodupon the sound spectra breaks through the limited contribution of the acousticmodality and outperforms the existing models' performance on both IEMOCAP andCMU-MOSEI datasets; (2) the introduction of the idea of multi-scale for visualextraction while single-branch for inference brings higher efficiency andmaintains the prediction accuracy at the same time; (3) the further integrationof data pre-processing into the aligned multimodal learning model allows thesignificant reduction of computational costs and storage space.

FV2ES: A Fully End2End Multimodal System for Fast Yet Effective Video Emotion Recognition Inference