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

Learning to capture long-range relations is fundamental to image/videorecognition. Existing CNN models generally rely on increasing depth to modelsuch relations which is highly inefficient. In this work, we propose the"double attention block", a novel component that aggregates and propagatesinformative global features from the entire spatio-temporal space of inputimages/videos, enabling subsequent convolution layers to access features fromthe entire space efficiently. The component is designed with a double attentionmechanism in two steps, where the first step gathers features from the entirespace into a compact set through second-order attention pooling and the secondstep adaptively selects and distributes features to each location via anotherattention. The proposed double attention block is easy to adopt and can beplugged into existing deep neural networks conveniently. We conduct extensiveablation studies and experiments on both image and video recognition tasks forevaluating its performance. On the image recognition task, a ResNet-50 equippedwith our double attention blocks outperforms a much larger ResNet-152architecture on ImageNet-1k dataset with over 40% less the number of parametersand less FLOPs. On the action recognition task, our proposed model achieves thestate-of-the-art results on the Kinetics and UCF-101 datasets withsignificantly higher efficiency than recent works.

$A^2$-Nets: Double Attention Networks