Kinship verification via ear images: A comparative study

Kinship verification using biometric traits is crucial for finding missing children, rapid forensic identification, and social media analysis. Ear biometrics is gaining attention due to its uniqueness, permanence, and non-intrusiveness. However, current research on ear-based kinship verification is limited, and the factors associated with ear similarities across kinship-related individuals require to be evaluated. To fill this gap, our study developed deep learning models to quantify the similarity between ear images and conduct the kinship verification task. Two ear image datasets, namely SCED and CNKE, were collected from Chinese subjects. The SimiNet model, which utilized a pre-trained ResNet50 as its backbone, was constructed to evaluate the cosine similarity between ear image pairs. For kinship verification, the VTrans model was established by combining a pre-trained VGG16 with Transformer modules. Based on the cosine similarity, the SimiNet model obtained 93.53% accuracy and an AUC (Area Under the Curve) of 0.98 for personal verification. The similarity analysis further revealed that kinship-related individuals of the same sex displayed higher ear similarity scores. The VTrans model attained 71.17% accuracy and an AUC of 0.76 on the CNKE dataset. Heatmaps revealed that the VTrans model focused on the helix and the upper half of the ear during kinship verification. The model code has been provided in Github ( https://anonymous.4open.science/r/SimNet_VTrans-EB41 ) to facilitate refinement in future research. Our study has successfully established deep learning models for ear similarity quantification and kinship verification, providing effective tools for biometric forensic identification.

The analysis of kin relations from visual data represents a challenging research problem with important real-world applications. However, research in this area has mostly been limited to the analysis of facial images, despite the potential of other physical (human) characteristics for this task. In this paper, we therefore study the problem of kinship verification from ear images and investigate whether salient appearance characteristics, useful for this task, can be extracted from ear data. To facilitate the study, we introduce a novel dataset, called KinEar, that contains data from 19 families with each family member having from 15 to 31 ear images. Using the KinEar data, we conduct experiments using a Siamese training setup and 5 recent deep learning backbones. The results of our experiments suggests that ear images represent a viable alternative to other modalities for kinship verification, as 4 out of 5 considered models reach a performance of over 60% in terms of the Area Under the Receiver Operating Characteristics (ROC-AUC).

Previous works show that human ears can be used for identification, gender classification, and kinship verification and have investigated whether there is a symmetry between a person's ears; however, the symmetry performances have been less than satisfactory. Our paper extends the analysis of gender classification on ear images and analyses bilateral symmetry of human ears, in both cases aiming to determine the ear parts from which recognition is derived. We use model-based analysis and deep learning methods to capitalize on structure and performance, respectively. We consider the rotation of ear images under an affine transformation, by modelling the ear as a flat plane attached to the head. We address the question as to whether it is possible that given an image of one ear, a person can then be recognized from their other ear. Such a symmetry-based strategy could reduce constraints on applications of ear biometrics. We show that it is possible to recognise the gender with a 90.9% success rate and that the ear rim (the upper helix and lobe) dominates performance. To investigate symmetry, we compare one ear with a mirrored version of the other ear and achieve 93.1% CCR, which is the current state-of-the-art, with important regions different from those determined for gender. To extend the analysis we construct two groups of images, one of which contains both ears from the same subject and the other contains two ears from different subjects. The 100% CCR confirms the existence of symmetry between a subject's ears. By these approaches we show that there is actually a high chance that there exists symmetry between a person's ears and that it would be prudent for recognition systems to concentrate on the inner ear rather than the outer ear.

Kinship Verification via facial images is an emerging research topic in the field of biometrics, pattern recognition, and computer vision. It is motivated by the findings that individuals with some genetic relations have certain similarities in their facial appearances. These similarities in the facial appearance is a result of inherited facial features from one generation to next generation, especially from parents to children. The researchers use these inherited facial features to perform Kinship Verification and validate their results. Most of the existing methods in Kinship Verification are based on metric learning which aims to improve the verification rate ignoring the effect of salient facial features. This paper aims to learn the effect of extracted facial features between pair of facial images to perform Kinship Verification. This paper proposes different feature descriptors to describe salient facial features and support vector machine classifier to learn these extracted facial features. To validate the accuracy of the proposed methods, experiments are performed on KinFaceW-I dataset. The results obtained outperformed previous results and would encourage researchers to focus on this emerging topic.

Kinship verification refers to finding a blood relationship between people. Facial images can be used to establish the relationship using feature extraction and matching. Kinship verification is an emerging topic for research in computer vision. The main task of kinship verification is to determine kin relation from a given pair of facial images. It is a binary classification problem that answers yes or no based upon whether the relationship exists or not for a given pair of images. The principle behind kinship verification is that every child inherits some of his parent's facial features. It has many real-world applications like finding missing children, image annotation, automatic photo management, and social media image tagging, etc. Convolutional neural networks have been used recently for classification problems in digital images. The major research contributions by CNN in solving kinship verification problems have been analyzed in this chapter.

Deep discriminant generation-shared feature learning for image-based kinship verification

This paper proposes a multi-view geometric mean metric learning (MvGMML) method for the real-world kinship verification from facial images. Unlike existing kinship verification methods which dramatically degrade their performance when facial images are not well aligned, we present an efficient misalignment-robust kinship verification framework. First, a facial feature detector is employed to localize several facial feature points such as the right and left corners of two eyes. Then, a dense SIFT descriptor is extracted around each feature point. Lastly, our proposed MvGMML method jointly learns multiple local geometric mean metrics, one geometric mean metric for each view (i.e., feature point), to better exploit complementary information of all views. Experimental results on two widely used kinship datasets are presented to show the efficacy of our method.

There are always two main problems in identification of human beings through their ear images: 1- If distances of the individual from camera changes, the sizes of ears in the saved images are varied in proportion to this distance. 2- If head of people in taken images is tilted upwards or downwards, this causes ear images of these people rotate in proportion to saved ear images in database. In both of these cases, all identification systems do not work properly. In this article, we proposed a new method for normalizing human ear images by detecting the rotation and scaling variation, and normalizing the ear images accordingly. Our proposed method works well on all ear databases and all ear images (either left or right) which have been taken from front side of the ears. Our method provides high performance to the biometric identification systems to identify human being, even when the images of human ears are taken from long distance with small scale.

Based on two facial image appearances estimating their kinship is the main aim of the kinship verification. Age progression-based kinship verification is one of the obscure parts in this research. The similarities in facial features between parent and their children will be numerous in their childhood. As age progress, child facial features are varied and dispersed from parent facial features. It becomes a challenging task to estimate their kinship. So, a new dimensional database with parent in childhood and their child images is collected. This paper proposes and trains a metric to ensure that the model can predict whether the given pair images are kin or non-kin. In training module, differences of Histogram of Gradient (HoG) features for all combinations of pairs are computed and each pair absolute differences are calculated. Further, selective minimum variances are used to assess the kin similarity features. A global threshold is computed to classify kins and non-kins. After this comprehensive training, testing is also done in a similar way. The computed global threshold in training module is effectively used to estimate kinship verification in testing module. Experimental results are presented and out performed with an accuracy of 82%.

Kinship verification from facial images in wild conditions is a relatively new and challenging problem in face analysis. Several datasets and algorithms have been proposed in recent years. However, most existing datasets are of small sizes and one standard evaluation protocol is still lack so that it is difficult to compare the performance of different kinship verification methods. In this paper, we present the Kinship Verification in the Wild Competition: the first kinship verification competition which is held in conjunction with the International Joint Conference on Biometrics 2014, Clearwater, Florida, USA. The key goal of this competition is to compare the performance of different methods on a new-collected dataset with the same evaluation protocol and develop the first standardized benchmark for kinship verification in the wild.

In this survey, kinship verification is defined as the automatic process of verifying whether two or more persons are blood relatives (kin) by analyzing images of their faces. Kinship verification is an important research field in computer vision with many applications such as finding missing persons, family album organization, and online image search. Although substantial progress has been made in kinship verification in the past decade, there are still challenges such as intrinsic (face i.e., differences in facial appearance) and extrinsic (acquisition i.e., varying imaging conditions) problems. And there is still a demand for more diverse datasets.Therefore, this paper provides a survey on kinship verification methods and datasets. The survey starts with the definition of kinship verification and its corresponding intrinsic and extrinsic challenges. Then, an overview of kinship verification methods and datasets is given. Finally, a new multi-modal dataset (Nemo-Kinship Dataset) is proposed as a benchmark dataset addressing large inter-subject age variations consisting of 4216 videos of 248 persons from 85 families. The newly collected dataset is used to systematically test and analyze state-of-the-art methods.

Kinship verification is the process of using observable characteristics, including facial features or other biometric data, to ascertain whether two people are familial related. Current approaches to kinship verification encounter multiple challenges, despite their significance in fields such as biometric security, family reunification, and forensic investigations. The most accurate way to confirm kinship is by DNA analysis, which is expensive, time-consuming, and invasive; in contrast, the traditional methods using manual comparison and facial recognition are more prone to human mistake. Visual verification of kinship among family members can be difficult to reliably rely on due to genetic variability, aging, and environmental influences. Automated methods that can reliably and efficiently confirm kinship using non-invasive methods are required, especially in situations where immediate identification is essential, like in the investigation of missing persons cases and disaster response. This study explores the use of different verification kinship techniques, ranging from various machine learning techniques reaching to up-to-date deep learning techniques. Finally, this study will discuss the datasets, preprocessing, methodology, advantages, disadvantages, and performance measures of selected state-of-the-art studies of kinship verification, and present the best technique used in kinship verification with the most related dataset. Index Terms— Kinship verification, Deep learning, Machine learning.

Features extracted from facial images are used in various fields such as kinship verification. The kinship verification system determines the kin or non-kin relation between a pair of facial images by analysing their facial features. In this research, different texture and color features have been used along with the metric learning method, to verify the kinship for the four kinship relations of father-son, father-daughter, mother-son and mother-daughter. First, by fusing effective features, NRML metric learning used to generate the discriminative feature vector, then SVM classifier used to verify to kinship relations. To measure the accuracy of the proposed method, KinFaceW-I and KinFaceW-II databases have been used. The results of the evaluations show that the feature fusion and NRML metric learning methods have been able to improve the performance of the kinship verification system. In addition to the proposed approach, the effect of feature extraction from the image blocks or the whole image is investigated and the results are presented. The results indicate that feature extraction in block form, can be effective in improving the final accuracy of kinship verification.

Faces is a unique region in our body that can be used as a biometric identity. Furthermore, the face between two people that have a kinship relationship may share the same face features which can be used to decide whether two people have a kinship relationship or not. In this paper, we proposed a family-aware convolutional neural network (CNN) for the visual kinship verification problem. Our proposed classifier is constructed by paralleling the state-of-the-art face recognition model and attaching two additional networks, a family-aware network, and a kinship verification network. The family-aware network weights adjusted by learning features specific to the family using deep metric learning loss while the kinship verification network use softmax loss to learn the kinship verification problem. One of the advantages of our proposed classifier is that the output of the classifier is normalized and can be represented as the probability of two images being kin or non-kin. To preserve the face recognition features extraction ability in the state-of-the-art face recognition model, we freeze the weights of the convolutional layers in the classifier for the training process. In the testing process, the family-aware network is detached to construct the final classifier. Experiments on FIW (Families In the Wild) dataset show that our proposed classifier performs better comparing with classifiers that trained without a family-aware network and the ensemble version of the classifier is comparable with several state-of-the-art methods with an average accuracy of 68.84%.

Kinship verification is a very important technique in many real-world applications, e.g., personal album organization, missing person investigation and forensic analysis. However, it is extremely difficult to verify a family pair with generation gap, e.g., father and son, since there exist both age gap and identity variation. It is essential to well fight off such challenges to achieve promising kinship verification performance. To this end, we propose a towards-young cross-generation model for effective kinship verification by mitigating both age and identity divergences. Specifically, we explore a conditional generative model to bring in an intermediate domain to bridge each pair. Thus, we could extract more effective features through deep architectures with a newly-designed Sparse Discriminative Metric Loss (SDM-Loss), which is exploited to involve the positive and negative information. Experimental results on kinship benchmark demonstrate the superiority of our proposed model by comparing with the state-of-the-art kinship verification methods.