Abstract:We have advanced an effort to develop vision based human understanding technologies for realizing human-friendly machine interfaces. Visual information, such as gender, age ethnicity, and facial expression play an important role in face-to-face communication. This paper addresses a novel approach for ethnicity classification with facial images. In this approach, the Gabor wavelets transformation and retina sampling are combined to extract key facial features, and support vector machines that are used for ethnicity classification. Our system, based on this approach, has achieved approximately 94% for ethnicity estimation under various lighting conditions.
1. Introduction
When we communicate directly with other people, visual information plays an important role. When we look at a person's face, we not only discern who it is, but also process other information about the person, such as emotion, gestures, ethnicity, age and gender, shape of the eyes and nose, and charm. Then, based on this information, we consciously or subconsciously adjust our interaction, such as speaking louder with elderly people, or switching language when speaking with foreigners. A face contains a great deal of information about that person; communication can flow more freely by using the sense of sight to understand this information. Like people communicating with each other, if machines were able to visually recognize and comprehend human faces, man-machine communication could flow more freely. If machines were to understand this visual information, it could be possible to create machines that could be operated safely, securely, simply and comfortably. Or, depending on a person's personal attributes, moods, preferences, or abilities, machines could adjust themselves to provide appropriate service interfaces, value added services, or information. To enable the adaptation of machines to the needs of humans, we aim to build technology for understanding people through visual information. We have already built face recognition technology and expanded face recognition technology through gender and age estimation technology. In this paper, we propose an automatic ethnicity estimation technology based on sampling gabor features from the eye and mouth regions and apply support vector machine to estimation. Also, we will present ethnicity-estimation algorithms in detail and discuss the experimental results.