HAL will be down for maintenance from Friday, June 10 at 4pm through Monday, June 13 at 9am. More information
Skip to Main content Skip to Navigation
Conference papers

Geometric Deep Neural Network using Rigid and Non-Rigid Transformations for Human Action Recognition

Abstract : Deep Learning architectures, albeit successful in most computer vision tasks, were designed for data with an underlying Euclidean structure, which is not usually fulfilled since pre-processed data may lie on a non-linear space. In this paper, we propose a geometry aware deep learning approach using rigid and non rigid transformation optimization for skeleton-based action recognition. Skeleton sequences are first modeled as trajectories on Kendall's shape space and then mapped to the linear tangent space. The resulting structured data are then fed to a deep learning architecture, which includes a layer that optimizes over rigid and non rigid transformations of the 3D skeletons, followed by a CNN-LSTM network. The assessment on two large scale skeleton datasets, namely NTU-RGB+D and NTU-RGB+D 120, has proven that the proposed approach outperforms existing geometric deep learning methods and exceeds recently published approaches with respect to the majority of configurations.
Complete list of metadata

Contributor : Hassen Drira Connect in order to contact the contributor
Submitted on : Friday, November 26, 2021 - 9:51:19 AM
Last modification on : Thursday, March 24, 2022 - 3:42:47 AM
Long-term archiving on: : Sunday, February 27, 2022 - 6:18:21 PM


Files produced by the author(s)


  • HAL Id : halshs-03450533, version 1


Rasha Friji, Hassen Drira, Faten Chaieb, Hamza Kchok, Sebastian Kurtek. Geometric Deep Neural Network using Rigid and Non-Rigid Transformations for Human Action Recognition. International Conference in Computer Vision, Oct 2021, Visio, France. ⟨halshs-03450533⟩



Record views


Files downloads