FaceQnet: Quality Assessment for Face Recognition based on Deep Learning

Abstract

In this paper we develop a Quality Assessment approach for face recognition based on deep learning. The method consists of a Convolutional Neural Network, FaceQnet, that is used to predict the suitability of a specific input image for face recognition purposes. The training of FaceQnet is done using the VGGFace2 database. We employ the BioLab-ICAO framework for labeling the VGGFace2 images with quality information related to their ICAO compliance level. The groundtruth quality labels are obtained using FaceNet to generate comparison scores. We employ the groundtruth data to fine-tune a ResNet-based CNN, making it capable of returning a numerical quality measure for each input image. Finally, we verify if the FaceQnet scores are suitable to predict the expected performance when employing a specific image for face recognition with a COTS face recognition system. Several conclusions can be drawn from this work, most notably: 1) we managed to employ an existing ICAO compliance framework and a pretrained CNN to automatically label data with quality information, 2) we trained FaceQnet for quality estimation by fine-tuning a pre-trained face recognition network (ResNet-50), and 3) we have shown that the predictions from FaceQnet are highly correlated with the face recognition accuracy of a state-of-the-art commercial system not used during development. FaceQnet is publicly available in GitHub 1 .

Makeup Presentation Attacks: Review and Detection Performance Benchmark

“…To obtain this subset, the CelebA face database has been filtered to only contain face images with heavy use of makeup, frontal pose, and closed mouth. Face sample quality assurance has been conducted using the FaceQNet algorithm [68], resulting in a total number of 641 face images of different subjects. Example images of the resulting subset of the CelebA face database are depicted in figure 10.…”

“…blur. In addition, the deep learning-based face image quality assessment algorithm FaceQNet [68] is employed. While the suitability of additional recent CNN-based general face PAD may theoretically be considered for the M-PAD task (recall section II), those systems utilise additional information sources (depth and/or near-infrared images) which are not available for any of the publicly available M-PA datasets; conversely, the aforementioned systems do not offer pre-trained models for RGB image data alone.…”

Double Ghost Convolution Attention Mechanism Network: A Framework for Hyperspectral Reconstruction of a Single RGB Image

“…It combines imaging innovation with spectral technology to enable the detection of the two-dimensional geometric space and spectral information of the target. Hyperspectral imaging uses this type of approach to generate high-resolution continuous narrow-band image data [ 1 ]. Hyperspectral images combine the image and spectral information of samples.…”

“…At a certain wavelength, the image will reflect a certain defect more significantly because different components have different spectral absorptions. The spectral information can fully reflect the differences in the physical structure and chemical composition of the sample, and it has therefore been widely used in face recognition [ 1 ], image classification [ 2 ], image recognition [ 3 ], image restoration [ 4 ], and many other applications. However, hyperspectral imaging equipment is expensive, complex, and difficult to move, which limits the further development of hyperspectral imaging research.…”

Improving the Neural Network Algorithm for Assessing the Quality of Facial Images

“…Many of the recent works consider the problem of face quality assessment from a different point of view: as an indicator that reflects the usefulness of the image for the specific algorithm being used. Algorithms for obtaining this value use one of the existing face recognition systems, based on which either the training and test dataset is marked up [6] or the finished model is obtained directly [7], [8]. The main idea is that the confidence of the face recognition system for a pair of images of the same person and the difference in the quality of these images are interrelated.…”