Robust Face Frontalization For Visual Speech Recognition - Inria - Institut national de recherche en sciences et technologies du numérique Access content directly
Conference Papers Year : 2021

Robust Face Frontalization For Visual Speech Recognition


Face frontalization consists of synthesizing a frontallyviewed face from an arbitrarily-viewed one. The main contribution of this paper is a robust frontalization method that preserves non-rigid facial deformations, i.e. expressions, to perform lip reading. The method iteratively estimates the rigid transformation (scale, rotation, and translation) and the non-rigid deformation between 3D landmarks extracted from an arbitrarily-viewed face, and 3D vertices parameterized by a deformable shape model. An important merit of the method is its ability to deal with large Gaussian and non-Gaussian errors in the data. For that purpose, we use the generalized Student-t distribution. The associated EM algorithm assigns a weight to each observed landmark, the higher the weight the more important the landmark, thus favoring landmarks that are only affected by rigid head movements. We propose to use the zero-mean normalized cross-correlation (ZNCC) score to evaluate the ability to preserve facial expressions. We show that the method, when incorporated into a deep lipreading pipeline, considerably improves the word classification score on an in-the-wild benchmark.
Fichier principal
Vignette du fichier
Kang-ICCV21W-HAL.pdf (4.58 Mo) Télécharger le fichier
Origin Files produced by the author(s)

Dates and versions

hal-03326002 , version 1 (25-08-2021)
hal-03326002 , version 2 (03-09-2021)
hal-03326002 , version 3 (08-11-2021)



Zhiqi Kang, Radu Horaud, Mostafa Sadeghi. Robust Face Frontalization For Visual Speech Recognition. ICCVW 2021 - International Conference on Computer Vision Workshops, IEEE, Oct 2021, Montreal - Virtual, Canada. pp.2485-2495, ⟨10.1109/ICCVW54120.2021.00281⟩. ⟨hal-03326002v3⟩
189 View
357 Download



Gmail Mastodon Facebook X LinkedIn More