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

Egocentric 3D human pose estimation (HPE) from images is challenging due tosevere self-occlusions and strong distortion introduced by the fish-eye viewfrom the head mounted camera. Although existing works use intermediateheatmap-based representations to counter distortion with some success,addressing self-occlusion remains an open problem. In this work, we leverageinformation from past frames to guide our self-attention-based 3D HPEestimation procedure -- Ego-STAN. Specifically, we build a spatio-temporalTransformer model that attends to semantically rich convolutional neuralnetwork-based feature maps. We also propose feature map tokens: a new set oflearnable parameters to attend to these feature maps. Finally, we demonstrateEgo-STAN's superior performance on the xR-EgoPose dataset where it achieves a30.6% improvement on the overall mean per-joint position error, while leadingto a 22% drop in parameters compared to the state-of-the-art.

Building Spatio-temporal Transformers for Egocentric 3D Pose Estimation