Deep Entwined Learning Head Pose and Face Alignment Inside an Attentional Cascade with Doubly-Conditional fusion - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2021

Deep Entwined Learning Head Pose and Face Alignment Inside an Attentional Cascade with Doubly-Conditional fusion

Arnaud Dapogny
  • Fonction : Auteur
Matthieu Cord

Résumé

Head pose estimation and face alignment constitute a backbone preprocessing for many applications relying on face analysis. While both are closely related tasks, they are generally addressed separately, e.g. by deducing the head pose from the landmark locations. In this paper, we propose to entwine face alignment and head pose tasks inside an attentional cascade. This cascade uses a geometry transfer network for integrating heterogeneous annotations to enhance landmark localization accuracy. Furthermore, we propose a doubly-conditional fusion scheme to select relevant feature maps, and regions thereof, based on a current head pose and landmark localization estimate. We empirically show the benefit of entwining head pose and landmark localization objectives inside our architecture, and that the proposed AC-DC model enhances the state-of-the-art accuracy on multiple databases for both face alignment and head pose estimation tasks.

Dates et versions

hal-03174162 , version 1 (18-03-2021)

Identifiants

Citer

Arnaud Dapogny, Kevin Bailly, Matthieu Cord. Deep Entwined Learning Head Pose and Face Alignment Inside an Attentional Cascade with Doubly-Conditional fusion. 2020 15th IEEE International Conference on Automatic Face and Gesture Recognition (FG 2020), Nov 2020, Buenos Aires, Argentina. pp.192-198, ⟨10.1109/FG47880.2020.00038⟩. ⟨hal-03174162⟩
44 Consultations
0 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More