Skip to Main content Skip to Navigation

Prédire l'âge de personnes à partir de photos du visage : une étude fondée sur la caractérisation et l'analyse de signes du vieillissement

Abstract : Age has always been an important identity attribute. Mankind has developed, throw the evolutionary process, an innate ability to classify individuals according to their ages. This classification is partly driven by the face and the anatomical transformations that occur with time. Many dermatological, cosmetic or surgical procedures are developed to fight again signs of aging. Therefore, we can wonder how these procedures modify the perceived age of people who achieve them. In order to build an algorithm that will predict someone age from his front face picture, we have study the signs of facial aging and their incidence on perceived age. Firstly we have analyzed the anatomical transformations that alter the adult face. Secondly, we have determined the signs of aging which mostly drive the human perception of age. Finally, we have built and validated a predictive model of age thanks to the results of the first two steps. Anatomical transformation of the face with age: The importance of 21 signs of aging has been measured on a representative panel of Caucasian women aged between 20 and 74 years. This data have enabled to build the kinetic of facial aging. Subjective judgment of age: The objective was to determine which signs drive observers' perception when evaluating people age. Forty height observers were therefore asked to give an age to the volunteers whose aging signs were previously measured. The perception of age was shown to be bias by the age and the gender of the observers. Moreover, the relation between the signs of aging and the perceived age was evaluated using Partial Least Square (PLS) regression. Particularly, it was shown that depending on the age or the gender of the graders, they do not similarly use the signs aging when predicting peoples' age. Age prediction model: Finally a model was proposed to predict peoples' age from their front face image using PLS regressions. The proposed model combines and the signs of aging related with the color, the shape and the texture of the face. Like the Active Appearance Model (AAM), the proposed model allows to strongly reduce the dimensionality of the information carried by the pixels values. However, this model is supervised and is thus appropriate in our context of learning by sample. The ability of our model to predict peoples' age is finally comparable with human.
Document type :
Complete list of metadata
Contributor : Theses Bupmc Connect in order to contact the contributor
Submitted on : Friday, April 12, 2013 - 4:23:05 PM
Last modification on : Tuesday, November 16, 2021 - 4:44:14 AM
Long-term archiving on: : Monday, April 3, 2017 - 4:43:32 AM


  • HAL Id : tel-00812718, version 1


Alex A. Nkengne. Prédire l'âge de personnes à partir de photos du visage : une étude fondée sur la caractérisation et l'analyse de signes du vieillissement. Ingénierie biomédicale. Université Pierre et Marie Curie - Paris VI, 2008. Français. ⟨NNT : 2008PA066082⟩. ⟨tel-00812718⟩



Record views


Files downloads