Sun. May 9th, 2021

Facial recognition has been more and more widely used recently; however, there are some issues in this field. One of them is facial makeup because it can change the facial appearance and compromise a biometric system. A recent study suggests a technique to improve facial recognition with makeup.

Image credit: kaboompics via Pixabay, free licence

It explores part-based representations. Different parts of a face are affected by cosmetics differently; therefore, this approach can increase the accuracy of face recognition. Two strategies of cropping the face are analyzed.

Firstly, splitting into four components: left periocular, including the eye and eyebrow, right periocular, nose, and mouth. Secondly, dividing the face into three facial thirds. After cropping, features are extracted using convolutional neural networks (CNN) and fused with the holistic score. The results show that this approach let to achieve improvements even without fine-tuning or retraining CNN models.

Recently, we have seen an increase in the global facial recognition market size. Despite significant advances in face recognition technology with the adoption of convolutional neural networks, there are still open challenges, as when there is makeup in the face. To address this challenge, we propose and evaluate the adoption of facial parts to fuse with current holistic representations. We propose two strategies of facial parts: one with four regions (left periocular, right periocular, nose and mouth) and another with three facial thirds (upper, middle and lower). Experimental results obtained in four public makeup face datasets and in a challenging cross-dataset protocol show that the fusion of deep features extracted of facial parts with holistic representation increases the accuracy of face verification systems and decreases the error rates, even without any retraining of the CNN models. Our proposed pipeline achieved state-of-the-art performance for the YMU dataset and competitive results for other three datasets (EMFD, FAM and M501).

Research paper: de Assis Angeloni, M. and Pedrini, H., “Improving Makeup Face Verification by Exploring Part-Based Representations”, arXiv:2101.07338. Link:

Source link