2018-00788 - Learning Morphologically Plausible Pose Transfer

Contrat renouvelable : Oui

Niveau de diplôme exigé : Bac + 5 ou équivalent

Fonction : Doctorant

A propos du centre ou de la direction fonctionnelle

Le centre de recherche Inria Grenoble Rhône-Alpes regroupe un peu moins de 800 personnes réparties au sein de 35 équipes de recherche et 9 services support à la recherche.

Ses effectifs sont distribués sur 5 campus à Grenoble et à Lyon, en lien étroit avec les laboratoires et les établissements de recherche et d'enseignement supérieur de Grenoble et Lyon, mais aussi avec les acteurs économiques de ces territoires.

Présent dans les domaines du logiciel, du calcul haute performance, de l'internet des objets, de l'image et des données, mais aussi de la simulation en océanographie et en biologie, il participe au meilleur niveau à la vie scientifique internationale par les résultats obtenus et les collaborations tant en Europe que dans le reste du monde.

Contexte et atouts du poste

The PhD is part of the AVATAR INRIA project, a collaborative project between several INRIA teams with the aim  to significantly advance the field of AVATAR modeling in particular by improving their realism. The PhD will be shared between the Mimetic team in Rennes, specialized in animation and the Morpheo team in Grenoble, specialized in moving shape capture.

Mission confiée

One of the objective of AVATAR is the ability to transfer the motion captured from a user to its avatar in a faithful way. A key aspect in this process is the ability to preserve incidence relationships,  e.g. contacts between body parts or with the environment, when animating  avatars.  As a result,  a body pose should not, in practice,  be limited to the traditional joint angle that model mainly the internal or anatomical pose but  should also account for external contextual information, such as relationships  in-between body surface points or with the environment. This is especially true with contacts between body parts that cannot be captured with joint angles only. In order to better model human pose, a set of works consider the “interaction mesh” [Ho10, Bernardin16], a graph structure that connects joint centers and can be used to preserve distances between these centers when transferring body poses to an avatar. Interaction graphs aim at capturing the contextual information linked to the motion. However, while better preserving the interaction between body parts, the interaction mesh is still unable to accurately capture and transfer body surface information. The purpose of this PhD is therefore to investigate innovative solutions that encode both  internal and external shape poses as well as recent deep learning strategies that can take benefit of existing body shape datasets to learn pose transfers.

 

Compétences

  • Master degree in Computer Science or Applied Mathematics
  • Creative and highly motivated

Avantages sociaux

  • Restauration subventionnée
  • Transports publics remboursés partiellement
  • Sécurité sociale
  • Congés payés
  • Aménagement du temps de travail
  • Installations sportives

Rémunération

1982€ the fisrt 2 years, then 2085€ the third year.