Art painting Data Collection

Abstract

The classification of Art painting images is a computer vision applications that is growing considerably. The goal of this technology, is to classify an art painting image automatically, in terms of artistic style, technique used, or its author. For this purpose, the image is analyzed extracting some visual features. Many articles related with these problems have been issued, but in general the proposed solutions are focused in a very specific field. In particular, algorithms are tested using images at different resolutions, acquired under different illumination conditions. That makes complicate the performance comparison of the different methods. In this context, it will be very interesting to construct a public art image database, in order to compare all the existing algorithms under the same conditions. This paper presents a large art image database, with their corresponding labels according to the following characteristics: title, author, style and technique. Furthermore, a tool that manages this database have been developed, and it can be used to extract different visual features for any selected image. This data can be exported to a file in CSV format, allowing researchers to analyze the data with other tools. During the data collection, the tool stores the elapsed time in the calculation. Thus, this tool also allows to compare the efficiency, in computation time, of different mathematical procedures for extracting image data.