K-FACE: A Large-Scale KIST Face Database in Consideration with Unconstrained Environments

TL;DR

The paper presents K-FACE, a large-scale, high-quality Korean face database built with a novel hemispherical capturing device (27 cameras, 10 lights) to provide 17.55 million images across 1,000 subjects. It ensures uniform age distribution (20s–50s) and near-equal gender representation, while labeling 27 poses, 35 lighting conditions, 3 expressions, and 6–7 accessories per subject, plus 2-light combos when wearing accessories. The dataset includes facial landmarks and bounding boxes, enabling precise analysis of how pose, illumination, expression, and occlusion affect recognition, aging, and 3D reconstruction tasks. K-FACE thus enables rigorous benchmarking and development of robust face analytics under highly varied real-world conditions, with demonstrated applicability to unconstrained face recognition, age estimation/aging simulation, and related modalities. The work also emphasizes manual data quality control and provides a structured data hierarchy to support configurable experimental environments.

Abstract

In this paper, we introduce a new large-scale face database from KIST, denoted as K-FACE, and describe a novel capturing device specifically designed to obtain the data. The K-FACE database contains more than 1 million high-quality images of 1,000 subjects selected by considering the ratio of gender and age groups. It includes a variety of attributes, including 27 poses, 35 lighting conditions, three expressions, and occlusions by the combination of five types of accessories. As the K-FACE database is systematically constructed through a hemispherical capturing system with elaborate lighting control and multiple cameras, it is possible to accurately analyze the effects of factors that cause performance degradation, such as poses, lighting changes, and accessories. We consider not only the balance of external environmental factors, such as pose and lighting, but also the balance of personal characteristics such as gender and age group. The gender ratio is the same, while the age groups of subjects are uniformly distributed from the 20s to 50s for both genders. The K-FACE database can be extensively utilized in various vision tasks, such as face recognition, face frontalization, illumination normalization, face age estimation, and three-dimensional face model generation. We expect systematic diversity and uniformity of the K-FACE database to promote these research fields.

Figures (8)

• Figure 1: Capturing device. (a) Concept design and (b) actual appearance.
• Figure 2: Positions of the cameras in the capturing device. (a) Vertical axis angle and (b) horizontal axis angle at 0° on the vertical axis.
• Figure 3: The positions of (a) facial landmarks and (b) bounding boxes.
• Figure 4: Structure of the K-FACE database.
• Figure 5: Sample images of accessories and expressions (neutral, smile, and disgust).