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 Tools annual

Global: Zendy Free Plan

Global: Zendy Tools monthly

Global: Zendy Plus monthly

Action recognition is a key algorithmic part of emerging on-the-edge smartvideo surveillance and security systems. Skeleton-based action recognition isan attractive approach which, instead of using RGB pixel data, relies on humanpose information to classify appropriate actions. However, existing algorithmsoften assume ideal conditions that are not representative of real-worldlimitations, such as noisy input, latency requirements, and edge resourceconstraints.  To address the limitations of existing approaches, this paper presentsReal-World Graph Convolution Networks (RW-GCNs), an architecture-level solutionfor meeting the domain constraints of Real World Skeleton-based ActionRecognition. Inspired by the presence of feedback connections in the humanvisual cortex, RW-GCNs leverage attentive feedback augmentation on existingnear state-of-the-art (SotA) Spatial-Temporal Graph Convolution Networks(ST-GCNs). The ST-GCNs' design choices are derived from informationtheory-centric principles to address both the spatial and temporal noisetypically encountered in end-to-end real-time and on-the-edge smart videosystems. Our results demonstrate RW-GCNs' ability to serve these applicationsby achieving a new SotA accuracy on the NTU-RGB-D-120 dataset at 94.1%, andachieving 32X less latency than baseline ST-GCN applications while stillachieving 90.4% accuracy on the Northwestern UCLA dataset in the presence ofspatial keypoint noise. RW-GCNs further show system scalability by running onthe 10X cost effective NVIDIA Jetson Nano (as opposed to NVIDIA Xavier NX),while still maintaining a respectful range of throughput (15.6 to 5.5 Actionsper Second) on the resource constrained device. The code is available here:https://github.com/TeCSAR-UNCC/RW-GCN.

Real-World Graph Convolution Networks (RW-GCNs) for Action Recognition in Smart Video Surveillance