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

Anticipating human actions is an important task that needs to be addressedfor the development of reliable intelligent agents, such as self-driving carsor robot assistants. While the ability to make future predictions with highaccuracy is crucial for designing the anticipation approaches, the speed atwhich the inference is performed is not less important. Methods that areaccurate but not sufficiently fast would introduce a high latency into thedecision process. Thus, this will increase the reaction time of the underlyingsystem. This poses a problem for domains such as autonomous driving, where thereaction time is crucial. In this work, we propose a simple and effectivemulti-modal architecture based on temporal convolutions. Our approach stacks ahierarchy of temporal convolutional layers and does not rely on recurrentlayers to ensure a fast prediction. We further introduce a multi-modal fusionmechanism that captures the pairwise interactions between RGB, flow, and objectmodalities. Results on two large-scale datasets of egocentric videos,EPIC-Kitchens-55 and EPIC-Kitchens-100, show that our approach achievescomparable performance to the state-of-the-art approaches while beingsignificantly faster.

Multi-Modal Temporal Convolutional Network for Anticipating Actions in Egocentric Videos