Classification of Modern Art Movements with Computational Methods: Initial Results

Altinoz O. T. , ÖZKAN ALTINÖZ M.

25th Signal Processing and Communications Applications Conference (SIU), Antalya, Türkiye, 15 - 18 Mayıs 2017 identifier identifier


In this study, the similarity between different art movements is investigated. For this purpose, five different art movements are selected and thirty paintings are determined from different painters. By using these paintings, the similarity between paintings inside art movements and from other modern art movements are shown and classified by using mathematical methods. Computational methods are used for classification. Bag of feature (BoF) and support vector machine (SVM) are utilized. Image features are extracted with BoF and these properties are classified to get histogram. These histogram data are applied to SVM for classification. In conclusion, modern art movements are classified, and based on these results the differences are discussed.