Self-portrait to ID Document face matching: CNN-Based face verification in cross-domain scenario

Abstract

Face verification approaches determine whether two given faces are from the same person. Recently, a new demand for face verification application which has become popular in commercial applications is the self-portrait and ID face matching, in which we compare the faces of a selfie shot by a subject and the face in a picture of her identification document. In this work, we proposed a novel approach for face verification in a cross-domain scenario, assuming we have only two images for each subject in the dataset. The method is based on siamese architecture with triplet-loss function. Experiments show the proposed model reaches good effectiveness for cross-domain face verification with low error rates, in comparison to other works of the literature.

Combination of texture feature extraction and forward selection for one-class support vector machine improvement in self-portrait classification

“…Self-validation is needed to facilitate the face verification process from identity documents and faces in real-time. The literature on selfie portraits and photo documents have been carried out by [1]- [4]. Most of them use the convolutional neural network (CNN) approach.…”