One major challenge in computer vision is to go beyond the modeling of individual objects and to investigate the bi-(one-versus-one) or tri-(one-versus-two) relationship among multiple visual entities, answering such questions as whether a child in a photo belongs to given parents. The child-parents relationship plays a core role in a family and understanding such kin relationship would have fundamental impact on the behavior of an artificial intelligent agent working in the human world. In this work, we tackle the problem of one-versus-two (tri-subject) kinship verification and our contributions are three folds: 1) a novel relative symmetric bilinear model (RSBM) introduced to model the similarity between the child and the parents, by incorporating the prior knowledge that a child may resemble a particular parent more than the other; 2) a spatially voted method for feature selection, which jointly selects the most discriminative features for the child-parents pair, while taking local spatial information into account; 3) a large scale tri-subject kinship database characterized by over 1,000 childparents families. Extensive experiments on KinFaceW, Family101 and our newly released kinship database show that the proposed method outperforms several previous state of the art methods, while could also be used to significantly boost the performance of one-versus-one kinship verification when the information about both parents are available.
Bi-subject kinship verification addresses the problem of verifying whether there exists some kind of kin relationship (i.e., father-son, father-daughter, motherson and mother-daughter) between a pair of parent-child subjects based purely on their visual appearance. The task is challenging due to the involvement of two different subjects possibly with different genders and ages. In addition, collecting sufficient training samples for each type of kinship is difficult. In this work, we present a novel method to address these issues by considering each type of kin relation verification as one task and learning them at one time in the framework of multi-task learning, by sharing feature sets and useful structures among the tasks.Particularly our contributions are three folds: first, we introduce a new type of learning problem, called mixed bi-subject kinship verification, to the topic of bi-subject kinship verification: instead of simply verifying whether some fixed kinship relationship (e.g., mother-son) can be established for a given pair of parent-child images, we try to figure out whether any type of the four kinship relations can be established according to the visual features of the image pair, with no need to know the genders of the subjects to be verified beforehand. Second, we propose a novel multi-task learning method to address this problem with two transformation matrices -one is shared amongst all the tasks and the other is unique to each task. Both matrices are simultaneously learned in a joint framework, which enables our algorithm to utilize the common knowledge of the four tasks. Third, we propose a multi-view multi-task learning(MMTL) method to perform multiple feature fusion to improve the mixed bi-subject kinship verification performance. Extensive experiments on the large scale KinFaceW kinship database demonstrate the feasibility and effectiveness of the proposed algorithm.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.