Face registration: evaluating generative models for automatic dense landmarking of the face

  • Authors:

  • Karl Ricanek;Amrutha Sethuram;Wankou Yang

  • Affiliations:

  • ISIS Institute - Face Aging Group, UNC Wilmington, Wilmington, NC;ISIS Institute - Face Aging Group, UNC Wilmington, Wilmington, NC;School of Automation, Southeast University, Nanjing, China

  • Venue:
  • IScIDE'12 Proceedings of the third Sino-foreign-interchange conference on Intelligent Science and Intelligent Data Engineering
  • Year:
  • 2012

Quantified Score

Hi-index 0.00

Visualization

Abstract

In this work we evaluate three generative techniques for automatic registration of more than 250 face landmarks (annotations). We compare/contrast these techniques based on developing general and a ethnic and gender specific models to detemine whether the specific, ethnic-gender, models can outperform the general model in accurately locating the dense landmarks. Further, we determine which of the three genrative tehcniques are more robust. The three techniques evaluted are the Active Shape Models (ASM), the Active Appearance Model (AAM), and the Constrained Local Model (CLM). In addition this work provides an understanding of the types of landmarks that each technique performs well on and the landmarks that the techniques perform poorly on. Further, it is shown that the performance of STASM and CLM are comparable and better than AAM and that specific models perform better than the general models.