E2E-GNet: An End-to-End Skeleton-based Geometric Deep Neural Network for Human Motion Recognition

Mubarak Olaoluwa, Hassen Drira · 2026

Read E2E-GNet: An End-to-End Skeleton-based Geometric Deep Neural Network for Human Motion Recognition by Mubarak Olaoluwa, Hassen Drira (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2603-02477.txt.

Description

Mubarak Olaoluwa and Hassen Drira recognize human motion from skeleton data in this 2026 paper. Skeleton sequences live in curved, non-Euclidean spaces, unlike flat image grids. Geometric deep learning is built for exactly such data. Their network, E2E-GNet, processes motion end to end in that curved setting.

A geometric transform layer sharpens the split between different motions. A special activation then projects the data onto a flat space for classification. Projection can distort skeleton shapes and blur key cues. A distortion-aware layer limits that damage during training. Tests across five data sets in three domains beat rival methods at lower cost. Ablation studies confirm each new layer pulls its weight.

Catalog record

CREATORMubarak Olaoluwa, Hassen Drira
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE49431
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2603.02477

Browse the full library · About the Tower of Babel