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

Deep learning has proved to be very effective in video action recognition.Video violence recognition attempts to learn the human multi-dynamic behavioursin more complex scenarios. In this work, we develop a method for video violencerecognition from a new perspective of skeleton points. Unlike the previousworks, we first formulate 3D skeleton point clouds from human skeletonsequences extracted from videos and then perform interaction learning on these3D skeleton point clouds. Specifically, we propose two types of Skeleton PointsInteraction Learning (SPIL) strategies: (i) Local-SPIL: by constructing aspecific weight distribution strategy between local regional points, Local-SPILaims to selectively focus on the most relevant parts of them based on theirfeatures and spatial-temporal position information. In order to capture diversetypes of relation information, a multi-head mechanism is designed to aggregatedifferent features from independent heads to jointly handle different types ofrelationships between points. (ii) Global-SPIL: to better learn and refine thefeatures of the unordered and unstructured skeleton points, Global-SPIL employsthe self-attention layer that operates directly on the sampled points, whichcan help to make the output more permutation-invariant and well-suited for ourtask. Extensive experimental results validate the effectiveness of our approachand show that our model outperforms the existing networks and achieves newstate-of-the-art performance on video violence datasets.

Improving Video Violence Recognition with Human Interaction Learning on 3D Skeleton Point Clouds