Makeup detection using Local Fisher Discriminant Analysis

Recent studies on face recognition, beauty analysis, age and health estimation, have found facial makeup application as a challenging issue. Hence, it is crucial for these algorithms to determine the makeup presence on faces. They can robust their recognition by utilizing makeup detection consequences. In this work we employ distinctive features to indicate the effect of makeup on the faces, moreover instead of utilizing whole face images, ten local patches have been extracted from images, then features assessed in each patch. Local patches which give the most information for each feature will be selected. At last two discriminant mapping methods, Marginal Fisher Analysis and Local Fisher Discriminant Analysis, are employed to trace more information from selected patches. Local Fisher Discriminant Analysis brought about accuracy rate of 95.03% for our makeup detection algorithm.