Multispectral Face Recognition Using Transfer Learning with Adaptation of Domain Specific Units

Abstract

Facial recognition is a method of identifying or authenticating the identity of people through their faces. Nowadays, facial recognition systems that use multispectral images achieve better results than those that use only visible spectral band images. In this work, a novel architecture for facial recognition that uses multiple deep convolutional neural networks and multispectral images is proposed. A domain-specific transfer-learning methodology applied to a deep neural network pre-trained in RGB images is shown to generalize well to the multispectral domain. We also propose a skin detector module for forgery detection. Several experiments were planned to assess the performance of our methods. First, we evaluate the performance of the forgery detection module using face masks and coverings of different materials. A second study was carried out with the objective of tuning the parameters of our domain-specific transfer-learning methodology, in particular which layers of the pre-trained network should be retrained to obtain good adaptation to multispectral images. A third study was conducted to evaluate the performance of support vector machines (SVM) and k-nearest neighbor classifiers using the embeddings obtained from the trained neural network. Finally, we compare the proposed method with other state-of-the-art approaches. The experimental results show performance improvements in the Tufts and CASIA NIR-VIS 2.0 multispectral databases, with a rank-1 score of 99.7% and 99.8%, respectively.

Faces in Event Streams (FES): An Annotated Face Dataset for Event Cameras

“…In most applications, face detection and facial landmark detection (see Figure 2 c) serve as the first mandatory steps. A substantial amount of research has been conducted in visual [ 22 , 23 , 24 ], thermal [ 25 , 26 ], depth [ 27 , 28 ], and other domains [ 29 ] for face and facial landmark detection. State-of-the-art face detection algorithms rely on deep learning networks [ 30 ], particularly convolutional neural networks (CNNs) [ 31 , 32 ], reinforcement learning [ 33 ], generative adversarial networks [ 34 ], and hybrid architectures [ 35 ].…”

Machine learning models in people detection and identification

“…By applying SVM, an efficiency of more than 90 % was achieved in the classification [36]. Likewise, Rayani and Rajakumar [37], used SVMs considering Additionally, Chambino et al [41], in their study of multispectral face recognition for classification, employed linear and radial basis support vector machines. The tests were run for two datasets with training epochs of 10 and 50 for each of these.…”

Joint Feature-Space and Sample-Space Based Heterogeneous Feature Transfer Method for Object Recognition Using Remote Sensing Images with Different Spatial Resolutions

“…To solve this problem, one can take into consideration the introduction of transfer learning technology [5,6], which refers to transfer knowledge or experience available for one or more domains (i.e., source domain) to improve the performance for a new yet related domain (i.e., target domain). The source domain or target domain consists of three concepts, feature space X , marginal probability distribution P(X), and conditional probability distribution P(Y|X), where X = {x 1 , x 2 , .…”