Construction and utilization of image database for image attribute structuring

Weon-Geun Oh, Seungjae Lee

Abstract: Image attribute structuring technology is a technology for analyzing and organizing the attributes of images and extracting various and valuable information included in the images by using them, and image database for developing and evaluating excellent technologies is an essential element. These image databases should be constructed in various forms according to the purpose of development technology and application field, and the evaluation architecture should also change. That is, the performance of the developed technology should be objectively evaluated and authenticated using a standardized experimental image database. As a result, technology developers can develop better image attribute structuring techniques and allow evaluators (or evaluation agencies) to evaluate them fairly. In this paper, we introduce various kinds of image databases and evaluation architectures that are used in international standards and competitions held in the field of computer vision recently.