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 Tools annual

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 Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

The goal of spatial-temporal action detection is to determine the time andplace where each person's action occurs in a video and classify thecorresponding action category. Most of the existing methods adoptfully-supervised learning, which requires a large amount of training data,making it very difficult to achieve zero-shot learning. In this paper, wepropose to utilize a pre-trained visual-language model to extract therepresentative image and text features, and model the relationship betweenthese features through different interaction modules to obtain the interactionfeature. In addition, we use this feature to prompt each label to obtain moreappropriate text features. Finally, we calculate the similarity between theinteraction feature and the text feature for each label to determine the actioncategory. Our experiments on J-HMDB and UCF101-24 datasets demonstrate that theproposed interaction module and prompting make the visual-language featuresbetter aligned, thus achieving excellent accuracy for zero-shot spatio-temporalaction detection. The code will be available athttps://github.com/webber2933/iCLIP.

Interaction-Aware Prompting for Zero-Shot Spatio-Temporal Action Detection