2D Landmark-Based Facial Asymmetry Assessment in the Clinical Case of Facial Paralysis

Benjamin Szczapa (1) , Mohamed Daoudi (1, 2) , Anis Kacem (1) , Pierre Guerreschi (3, 4) , Ludwig Gebert (5) , Juan Carlos Alvarez-Paiva (6)

1 CRIStAL - Centre de Recherche en Informatique, Signal et Automatique de Lille - UMR 9189
2 IMT Lille Douai - Ecole nationale supérieure Mines-Télécom Lille Douai
3 Université de Lille
4 Service de chirurgie plastique, esthétique et reconstructrice
5 ervice de chirurgie plastique, reconstructrice et esth\'{e}tique du Centre Hospitalier Saint-Joseph Saint-Luc
6 LPP - Laboratoire Paul Painlevé - UMR 8524

In this paper, we propose a novel technique for quantifying the facial asymmetry from 2D videos to evaluate facial paralysis treatments based on Botulinum Toxin (BT) injections. Our approach uses 2D facial landmarks and barycentric coordinates to objectively quantify the facial asymmetry across 2D videos. To assess our approach, a new dataset of 2D videos, containing eighteen patients before and after the treatments have been collected. For each patient, we have collected nine facial expressions. Experimental results on the newly collected dataset show that the proposed approach provides promising results in concordance with clinical annotations.

Domaines: Vision par ordinateur et reconnaissance de formes [cs.CV]

Benjamin Szczapa, Mohamed Daoudi, Anis Kacem, Pierre Guerreschi, Ludwig Gebert, et al.. 2D Landmark-Based Facial Asymmetry Assessment in the Clinical Case of Facial Paralysis. The 14th IEEE International Conference on Automatic Face and Gesture Recognition (FG 2019), May 2019, Lille, France. ⟨hal-02371137⟩