Efficient Face Verification Under Makeup Using Few Salient Facial Regions

EasyChair Preprint no. 3833

Abstract

Automatic recognition of persons has attracted the attention of many researchers during the last years due to its many applications in various fields. However, this task faces several challenges related to many changes that can affect the human face. In particular, make-up faces represent a major challenge for facial recognition and verification. To deal with this issue, we propose an efficient salient patch-based method for verifying faces under makeup variation. Firstly, we use Mutli-Task Cascaded Convolutional Neural Networks (MTCCNN) to jointly, detect and align the face with five landmarks. The Histogram of Oriented Gradients (HOG) descriptor and Local Binary Patterns (LBP) are then adopted to represent the face by concatenating their histogram features in few salient regions around the detected landmarks. Finally, we calculate the similarity measure between the extracted features to compare the two faces and determine whether they are for the same person or not. The performance of the proposed method is validated on the challenging Y MU (YouTube Makeup dataset ) and MIF S (Makeup Induced Face Spoofing) datasets. The obtained results proved the superiority of the proposed method against multipatch based method from the state of the art.

Keyphrases: Face verification, Histogram Oriented of Gradients, Local Binary Patterns, Mutli-task Cascaded Convolutional Networks

@Booklet{EasyChair:3833,
  author = {Khouloud Ferchichi and Haythem Ghazouani and Walid Barhoumi},
  title = {Efficient Face Verification Under Makeup Using Few Salient Facial Regions},
  howpublished = {EasyChair Preprint no. 3833},

  year = {EasyChair, 2020}}