AestheticNet: deep convolutional neural network for person identification from visual aesthetic
- PDF / 2,801,569 Bytes
- 11 Pages / 595.276 x 790.866 pts Page_size
- 39 Downloads / 202 Views
ORIGINAL ARTICLE
AestheticNet: deep convolutional neural network for person identification from visual aesthetic A. S. M. Hossain Bari1
· Brandon Sieu1 · Marina L. Gavrilova1
© Springer-Verlag GmbH Germany, part of Springer Nature 2020
Abstract A person’s visual aesthetics is an emerging behavioral biometric. Visual aesthetics can be defined as a person’s principles pertaining to their sense of beauty or fondness. Utilizing a person’s preference to certain images as discriminatory features forms the basis of person identification from visual aesthetics. This paper proposes a novel three-stage framework based on the convolutional neural network, AestheticNet, for the extraction of high-level features and identification of individuals from visual aesthetics. The rank-1 accuracy of 97.73% and rank-5 accuracy of 99.85% are achieved on the publicly available benchmark dataset, which outperforms all state-of-the-art methods. Keywords Visual aesthetic · Behavioral biometric · Deep learning · Principal component analysis · Residual learning-based convolutional neural network
1 Introduction The domain of biometric identification focuses on using discriminatory features solely belonging to a person. Biometrics based on individual’s interactions, habits, and conduct are called behavioral biometrics [28]. Within the domain of behavioral biometrics, social behavioral biometrics analyzes a person’s relations, communication, and attitudes [27]. Over the last decade, social networks have allowed for the widespread expression of these traits online. A person’s visual aesthetics is a prominent candidate for a social behavioral trait. Visual aesthetics can be described as an individual’s preferences that represent their sense of beauty or fondness toward images [20]. Given a set of user’s preferred images, constructing a visual aesthetic model for person identification is a pattern recognition problem. Due to the covert nature and accessibility of a person’s aesthetic profile online, this form of identification can be performed quickly and remotely. The advantage in accessibility can be
B
A. S. M. Hossain Bari [email protected] Brandon Sieu [email protected] Marina L. Gavrilova [email protected]
1
applied to multi-factor authentication, where a user’s aesthetic preferences are used in conjunction with other forms of identification. Visual aesthetic identification can also provide insights into the exploration of human preference and online social behavior. In the context of computer graphics, such a system could be extended for the automatic generation of personalized art [16] and user-adapted content in games [12]. If the pattern in an individual’s aesthetic preferences is known and learned, it is possible to then generate personally catered experiences with similar patterns. The ability for a system to extract and understand an individual’s aesthetics from example can be further extended to social robots by enhancing the areas of perception, personality, and inference [19]. By allowing a service robot t
Data Loading...
