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

Human Pose and Shape Estimation (HPSE) from RGB images can be broadlycategorized into two main groups: parametric and non-parametric approaches.Parametric techniques leverage a low-dimensional statistical body model forrealistic results, whereas recent non-parametric methods achieve higherprecision by directly regressing the 3D coordinates of the human body. Despitetheir strengths, both approaches face limitations: the parameters ofstatistical body models pose challenges as regression targets, and predicting3D coordinates introduces computational complexities and issues related tosmoothness. In this work, we take a novel approach to address the HPSE problem.We introduce a unique method involving a low-dimensional discrete latentrepresentation of the human mesh, framing HPSE as a classification task.Instead of predicting body model parameters or 3D vertex coordinates, our focusis on forecasting the proposed discrete latent representation, which can bedecoded into a registered human mesh. This innovative paradigm offers two keyadvantages: firstly, predicting a low-dimensional discrete representationconfines our predictions to the space of anthropomorphic poses and shapes;secondly, by framing the problem as a classification task, we can harness thediscriminative power inherent in neural networks. Our proposed model, VQ-HPS, atransformer-based architecture, forecasts the discrete latent representation ofthe mesh, trained through minimizing a cross-entropy loss. Our resultsdemonstrate that VQ-HPS outperforms the current state-of-the-art non-parametricapproaches while yielding results as realistic as those produced by parametricmethods. This highlights the significant potential of the classificationapproach for HPSE.

VQ-HPS: Human Pose and Shape Estimation in a Vector-Quantized Latent Space