Predicting human creativity perception

Creativity is a delicate subject when it comes to the understanding of how/when and why it occurs. In the last two decades considerable contributions incorporating creative mechanisms in computer applications were made. Recently, the discussion is turning towards the problems that are under the process of evaluating computer created visual objects, mainly in the area related to the automating aesthetic evaluation. This work aims to contribute to the understanding of the process of evaluating visual objects (with the help of human cognitive process of evaluation) in the specific area of visual arts, within computerized systems. By means of an empirical and automated approach we surveyed human subjects for the classification of the concept of creativity. Simultaneously, 804 features were extracted from around 2600 images used in order to model the human classification through an artificial neural network. This work proposes the use of classification models to understand the human creativity perception in a set of images. This is an important aspect not only to classify a product as being novel, but also to provide information about the creativity perception.