What is the Correct Hanging Orientation for Abstract Painting? Orientation Judgment and Detection

Ruyi Bai, Xiaoying Guo, Chunhua Jia

International Conference on Computer Science and Application Engineering (CSAE) · 2019 · 인용 5

Artists decide the orientation on which an abstract painting should be hung according to their inspiration, but for other amateur viewers, the correct hanging orientation that creates the highest aesthetic appreciation for an abstract painting is usually unknown. In this paper, we analyze the aesthetic evaluation of abstract painting from different hanging orientations through eye movement experiments. The experimental results show that abstract paintings with the correct hanging orientation usually get higher aesthetic evaluations.

However, how to determine the correct hanging orientation is still unclear. Therefore, we propose a novel systematic method to automatically detect the correct hanging orientation for abstract paintings. In this method, the global characteristics of paintings are firstly processed by CNN on the resized image.

Secondly, DNN is trained on 25 local features that are extracted according to the theories of abstract art. The results show that the accuracy rates of CNN and DNN are 59.3% and 68.1% respectively. Therefore, to increase the efficiency of classification, the combined model stacking the results (predicted probabilities) of CNN and DNN, uses SVM to predict the correct orientation of painting.

The results show that the combined model obtains an accuracy of 77.4%, outperforming each separate model in the state of the art. In addition, this work provides a new perspective of orientation detection for the aesthetic evaluation of abstract painting.

📄 이 논문을 인용한 Paperis 글

이 논문이 근거 목록에 올라 있는 Paperis 글입니다.