A Cross-age Kinship Verification Scheme Using Face Age Transfer Model

• Published in: 2022 IEEE International Conference on Big Data and Smart Computing (BigComp) pp. 206 - 210
• Main Authors: Kim, Hyeonwoo, Kim, Hyungjoon, Ko, Bumyeon, Hwang, Eenjun
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.01.2022
• Subjects: Computational modeling; Convolutional neural network; Data models; Deep learning; Face age transfer; Facial image analysis; Generative adversarial network; Genetics; Image analysis; Image processing; Kinship verification; Neural networks; Social networking (online)
• ISSN: 2375-9356
• DOI: 10.1109/BigComp54360.2022.00047

As the amount of face images collected from diverse devices such as smartphones, CCTVs and high-definition cameras is rapidly increasing, various face-based applications such as finding missing family members, social media analysis and genealogical research have attracted much attention. To effectively implement such applications, a technique called kinship verification can be used. Kinship verification determines the genetic relationship between people through facial image analysis. Recently, as convolutional neural network (CNN) shows good performance in image processing, many CCN-based kinship verification methods have been proposed. However, they suffered from problems such as insufficient labeled data to train deep learning models, and poor validation accuracy when the age difference between people was large. To alleviate these problems, in this paper, we propose a cross-age kinship verification scheme using a face age transfer model. To prove the effectiveness of the proposed scheme, we conducted several comparative experiments with other models, and we confirmed that the verification accuracy and f1 score of our proposed method improved.