Learning pairwise SVM on hierarchical deep features for ear recognition

Abstract

Convolutional neural networks (CNNs)-based deep features have been demonstrated with remarkable performance in various vision tasks, such as image classification and face verification. Compared with the hand-crafted descriptors, deep features exhibit more powerful representation ability. Typically, higher layer features contain more semantic information, while lower layer features can provide more low-level description. In addition, it turns out that the fusion of different layer features will lead to superior performance. Here, we propose a novel approach for human ear identification by combining hierarchical deep features. First, hierarchical deep features are extracted from ear images using CNN pre-trained on large-scale data set. To enhance the feature representation and reduce the high dimension of deep features, the discriminant correlation analysis (DCA) is adopted for fusing deep features from different layers for further improvement. Owing to the lack of ear images per person, the authors propose to transform the ear identification problem to the binary classification by composing pairwise samples and resolve it with the pairwise support vector machine (SVM). Experiments are conducted on four public databases: USTB I, USTB II, IIT Delhi I, and IIT Delhi II. The proposed method achieves promising recognition rate and exhibits decent performance compared with the state-of-the-art methods.

Interpol review of fingermarks and other body impressions 2016–2019

“…Similarly, the anatomy of the external ear received some attention by researchers [ [576] , [577] , [578] , [579] ], some insisting on the uniqueness of the organ [ 579 ] but the most promising lines of inquiry are linked to the use of the ear as a biometric system [ [580] , [581] , [582] , [583] , [584] ], including age and gender estimation [ 577 ]. We are not aware of research on earprints or earmarks, the most recent efforts being focused on the external organ.…”

Ensembles of Deep Learning Models and Transfer Learning for Ear Recognition

“…In earlier studies, multiple ear image characteristics including texture, edges, shape contours, and gradient information were utilized to better describe the ear features. For instance, texture descriptors have been extensively studied in ear recognition, justifying the importance of texture cues when featuring ear images [7,8,9]. On the other hand, methods exploiting gradient magnitude and orientation are also considered for a better description of the ear shape and contour [10,11,12].…”

Handcrafted versus CNN Features for Ear Recognition

“…Pattern recognition is now shifting from conventional handcrafted features to learned or CNN based image features [34][35][36]. Moreover, recent advancement has pushed the research area to study the recognition performance under more challenging conditions commonly referred to as unconstrained or in the wild.…”