Recent Advances in Deep Learning Techniques for Face Recognition

Abstract

In recent years, researchers have proposed many deep learning (DL) methods for various tasks, and particularly face recognition (FR) made an enormous leap using these techniques. Deep FR systems benefit from the hierarchical architecture of the DL methods to learn discriminative face representation. Therefore, DL techniques significantly improve state-of-the-art performance on FR systems and encourage diverse and efficient real-world applications. In this paper, we present a comprehensive analysis of various FR systems that leverage the different types of DL techniques, and for the study, we summarize 171 recent contributions from this area. We discuss the papers related to different algorithms, architectures, loss functions, activation functions, datasets, challenges, improvement ideas, current and future trends of DL-based FR systems. We provide a detailed discussion of various DL methods to understand the current state-of-the-art, and then we discuss various activation and loss functions for the methods. Additionally, we summarize different datasets used widely for FR tasks and discuss challenges related to illumination, expression, pose variations, and occlusion. Finally, we discuss improvement ideas, current and future trends of FR tasks. INDEX TERMSDeep learning, face recognition, artificial neural network, convolutional neural network, auto encoder, generative adversarial network, deep belief network, reinforcement learning.

The Face Inversion Effect in Deep Convolutional Neural Networks

“…Here we used a representative DCNN for face recognition, VGG-Face (Parkhi et al, 2015 ), which is pretrained to identify faces only. In recent years, various deep learning methods have been used in face recognition systems (Fuad et al, 2021 ). Among the various methods, DCNN is the most popular deep learning method for face recognition (Fuad et al, 2021 ).…”

“…In recent years, various deep learning methods have been used in face recognition systems (Fuad et al, 2021 ). Among the various methods, DCNN is the most popular deep learning method for face recognition (Fuad et al, 2021 ). Further, visual geometry group network-face (VGG-Face) is one of the most commonly used CNN models for face recognition (e.g., Ghazi and Ekenel, 2016 ; Karahan et al, 2016 ; Grm et al, 2017 ) and has shown successful performance of face recognition under various conditions (Ghazi and Ekenel, 2016 ).…”

Face Recognition With Masks Based on Spatial Fine-Grained Frequency Domain Broadening

“…The proposed method uses OpenCV to perform region of interest (ROI) cropping on facial images to recover an image of the non-occluded area. To avoid pixelation of the facial image after ROI processing, in which the image loses detail [13], we adopt an ESRGAN model [14] to enhance the details of the non-occluded image. ESRGAN can generate real textures and enhance detailed features through image super-resolution.…”

Deep Learning Applied to Intracranial Hemorrhage Detection

“…Deep Learning [13] is a series of models, technologies, and architectures based on Artificial Neural Networks which have undergone a revolution in AI in the last years. Its doubtless success in real world problems reaches areas such as face recognition [14], music composition [15] o multilingual translation [16]. In the last decade, thousands of researchers have proposed new models that overtake the achievements of previous architectures, so it is impossible to give a general framework that covers all possible approaches in deep learning.…”