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

Estimating 3D human motion from an egocentric video sequence plays a criticalrole in human behavior understanding and has various applications in VR/AR.However, naively learning a mapping between egocentric videos and human motionsis challenging, because the user's body is often unobserved by the front-facingcamera placed on the head of the user. In addition, collecting large-scale,high-quality datasets with paired egocentric videos and 3D human motionsrequires accurate motion capture devices, which often limit the variety ofscenes in the videos to lab-like environments. To eliminate the need for pairedegocentric video and human motions, we propose a new method, Ego-Body PoseEstimation via Ego-Head Pose Estimation (EgoEgo), which decomposes the probleminto two stages, connected by the head motion as an intermediaterepresentation. EgoEgo first integrates SLAM and a learning approach toestimate accurate head motion. Subsequently, leveraging the estimated head poseas input, EgoEgo utilizes conditional diffusion to generate multiple plausiblefull-body motions. This disentanglement of head and body pose eliminates theneed for training datasets with paired egocentric videos and 3D human motion,enabling us to leverage large-scale egocentric video datasets and motioncapture datasets separately. Moreover, for systematic benchmarking, we developa synthetic dataset, AMASS-Replica-Ego-Syn (ARES), with paired egocentricvideos and human motion. On both ARES and real data, our EgoEgo model performssignificantly better than the current state-of-the-art methods.

Ego-Body Pose Estimation via Ego-Head Pose Estimation