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

In multi-person 2D pose estimation, the bottom-up methods simultaneouslypredict poses for all persons, and unlike the top-down methods, do not rely onhuman detection. However, the SOTA bottom-up methods' accuracy is stillinferior compared to the existing top-down methods. This is due to thepredicted human poses being regressed based on the inconsistent human boundingbox center and the lack of human-scale normalization, leading to the predictedhuman poses being inaccurate and small-scale persons being missed. To push theenvelope of the bottom-up pose estimation, we firstly propose multi-scaletraining to enhance the network to handle scale variation with single-scaletesting, particularly for small-scale persons. Secondly, we introduce dualanatomical centers (i.e., head and body), where we can predict the human posesmore accurately and reliably, especially for small-scale persons. Moreover,existing bottom-up methods use multi-scale testing to boost the accuracy ofpose estimation at the price of multiple additional forward passes, whichweakens the efficiency of bottom-up methods, the core strength compared totop-down methods. By contrast, our multi-scale training enables the model topredict high-quality poses in a single forward pass (i.e., single-scaletesting). Our method achieves 38.4\% improvement on bounding box precision and39.1\% improvement on bounding box recall over the state of the art (SOTA) onthe challenging small-scale persons subset of COCO. For the human pose APevaluation, we achieve a new SOTA (71.0 AP) on the COCO test-dev set with thesingle-scale testing. We also achieve the top performance (40.3 AP) on OCHumandataset in cross-dataset evaluation.

Bottom-Up 2D Pose Estimation via Dual Anatomical Centers for Small-Scale Persons