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

Landmark-based human action recognition in videos is a challenging task incomputer vision. One key step is to design a generic approach that generatesdiscriminative features for the spatial structure and temporal dynamics. Tothis end, we regard the evolving landmark data as a high-dimensional path andapply non-linear path signature techniques to provide an expressive, robust,non-linear, and interpretable representation for the sequential events. We donot extract signature features from the raw path, rather we propose pathdisintegrations and path transformations as preprocessing steps. Pathdisintegrations turn a high-dimensional path linearly into a collection oflower-dimensional paths; some of these paths are in pose space while others aredefined over a multiscale collection of temporal intervals. Pathtransformations decorate the paths with additional coordinates in standard waysto allow the truncated signatures of transformed paths to expose additionalfeatures. For spatial representation, we apply the signature transform tovectorize the paths that arise out of pose disintegration, and for temporalrepresentation, we apply it again to describe this evolving vectorization.Finally, all the features are collected together to constitute the input vectorof a linear single-hidden-layer fully-connected network for classification.Experimental results on four datasets demonstrated that the proposed featureset with only a linear shallow network and Dropconnect is effective andachieves comparable state-of-the-art results to the advanced deep networks, andmeanwhile, is capable of interpretation.

Developing the Path Signature Methodology and its Application to Landmark-based Human Action Recognition