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/

Zendy Plus

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

Zendy Tools

Zendy Open

Human poses that are rare or unseen in a training set are challenging for anetwork to predict. Similar to the long-tailed distribution problem in visualrecognition, the small number of examples for such poses limits the ability ofnetworks to model them. Interestingly, local pose distributions suffer lessfrom the long-tail problem, i.e., local joint configurations within a rare posemay appear within other poses in the training set, making them less rare. Wepropose to take advantage of this fact for better generalization to rare andunseen poses. To be specific, our method splits the body into local regions andprocesses them in separate network branches, utilizing the property that ajoint position depends mainly on the joints within its local body region.Global coherence is maintained by recombining the global context from the restof the body into each branch as a low-dimensional vector. With the reduceddimensionality of less relevant body areas, the training set distributionwithin network branches more closely reflects the statistics of local posesinstead of global body poses, without sacrificing information important forjoint inference. The proposed split-and-recombine approach, called SRNet, canbe easily adapted to both single-image and temporal models, and it leads toappreciable improvements in the prediction of rare and unseen poses.

SRNet: Improving Generalization in 3D Human Pose Estimation with a Split-and-Recombine Approach