Abstract: Recent research has demonstrated that machine learning algorithms have understood individual face image fairly well. However, one major challenge in computer vision is to investigate the relationship among multiple visual entities. Indeed parents-child relationship plays a core role in a family and understanding such kind of kin relationship would have fundamental impact on the behavior of an artificial intelligent agent working in a human world. To address this issue, this paper collects by far the biggest kinship face database named LKFW (labeled KinFace in the wild) by searching public figures and family photos from the Internet. Then this paper presents a large margin bilinear classifier (LMBC) for kinship verification by effectively exploiting the dependence structure between child and parents in a few aspects: similarity measure and classifier design. In addition, the method can be solved as a quadratic programming problem and kernel is utilized. Extensive experiments on KinFaceW and LKFW kinship databases show that the proposed method can obtain better or comparable results.

Key words: kinship verification, bilinear classifier, face verification

摘要： 目前机器学习算法已经能够很好地理解单张人脸图像，但是探究多个主体之间的关系仍然是计算机视觉领域的一大挑战。父母-孩子关系是家庭关系的核心部件，研究这一关系对促进人工智能加深对人类社会的理解有重要影响。为此，通过在互联网上搜索名人及其家人的照片建立了一个目前最大的亲子关系图像数据库LKFW（labeled KinFace in the wild）。基于大间隔双线性分类器（large margin bilinear classifier，LMBC）进行亲子关系验证，其中包括从相似性度量、分类器设计这两个方面来探究孩子和父母之间的空间结构依赖关系。算法借助嵌入核的二次规划进行求解，在KinFaceW和LKFW亲子关系数据库上都达到了比最好性能更好或相当的性能。

关键词: 亲子验证, 双线性分类器, 人脸验证