Learning and Transfer of Deformations - Archive ouverte HAL Accéder directement au contenu
Rapport Année : 2019

Learning and Transfer of Deformations

C. Samir
H. Laga
  • Fonction : Auteur

Résumé

We propose a new framework for modeling surface dynamics, analyzing their elastic shapes, and predicting future dynamics from a given surface. We represent a facial expression, hereinafter referred to as a 4D facial surface, using its initial surface and a sequence of deformation vector fields. We then treat facial expressions as sampling from a mixture of spatio-temporal stochastic processes. This fact, in connection with expression's patterns learning, has made the incorporation of meaningful prior knowledge possible. The proposed framework can be applied to a wide range of applications where the analysis and the prediction of temporal shape dynamics are important. In this paper, we focus on 4D facial expressions and demonstrate the utility of the proposed framework in performing temporal alignment between facial expressions, deforming one 4D surface (\ie a facial expression) onto another, and simulating and synthesizing 4D surfaces.
Fichier principal
Vignette du fichier
Transfer-Expression-TechReport (1).pdf (10.21 Mo) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)

Dates et versions

hal-04518231 , version 1 (09-04-2024)

Identifiants

  • HAL Id : hal-04518231 , version 1

Citer

C. Samir, T. Deregnaucourt, H. Laga. Learning and Transfer of Deformations. Université Clermont Auvergne. 2019. ⟨hal-04518231⟩

Collections

PRES_CLERMONT LARA
0 Consultations
0 Téléchargements

Partager

Gmail Facebook X LinkedIn More