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

We propose a novel attention-based 2D-to-3D pose estimation network forgraph-structured data, named KOG-Transformer, and a 3D pose-to-shape estimationnetwork for hand data, named GASE-Net. Previous 3D pose estimation methods havefocused on various modifications to the graph convolution kernel, such asabandoning weight sharing or increasing the receptive field. Some of thesemethods employ attention-based non-local modules as auxiliary modules. In orderto better model the relationship between nodes in graph-structured data andfuse the information of different neighbor nodes in a differentiated way, wemake targeted modifications to the attention module and propose two modulesdesigned for graph-structured data, graph relative positional encodingmulti-head self-attention (GR-MSA) and K-order graph-oriented multi-headself-attention (KOG-MSA). By stacking GR-MSA and KOG-MSA, we propose a novelnetwork KOG-Transformer for 2D-to-3D pose estimation. Furthermore, we propose anetwork for shape estimation on hand data, called GraAttention shape estimationnetwork (GASE-Net), which takes a 3D pose as input and gradually models theshape of the hand from sparse to dense. We have empirically shown thesuperiority of KOG-Transformer through extensive experiments. Experimentalresults show that KOG-Transformer significantly outperforms the previousstate-of-the-art methods on the benchmark dataset Human3.6M. We evaluate theeffect of GASE-Net on two public available hand datasets, ObMan andInterHand2.6M. GASE-Net can predict the corresponding shape for input pose withstrong generalization ability.

K-Order Graph-oriented Transformer with GraAttention for 3D Pose and Shape Estimation