Deep Correlation Features for Image Style Classification

Wei-Ta Chu and Yi-Ling Wu

Multimedia Computing Laboratory
Dept. of Computer Science and Information Engineering
National Chung Cheng University

1. Introduction

This paper presents a comprehensive study of deep correlation features on image style classification. Inspired by that correlation between feature maps can effectively describe image texture, we design and transform various such correlations into style vectors, and investigate classification performance brought by different variants. In addition to intra-layer correlation, we also propose inter-layer correlation and verify its benefit. Through extensive experiments on image style classification and artist classification, we demonstrate that the proposed style vectors significantly outperforms CNN features coming from fully-connected layers, as well as outperforms the state-of-the-art deep representation.

2. The OilPainting Dataset [752 MB] (Link)

The OilPainting dataset consists of 17 painting styles and includes 19,797 images in total.

3. The OilPainting Artist Dataset [592 MB] (Link)

The OilPainting Artist dataset consists of artworks produced by 104 artists, and includes 15,357 images in total.

3. Citation

Please cite our work if you utilize this dataset.

Wei-Ta Chu and Yi-Ling Wu, "Deep Correlation Features for Image Style Classification," Proceedings of ACM Multimedia, pp. 402-406, 2016.


Any problem please contact .

[Main Page]

Last Updated: October 15, 2017