Home // SIGNAL 2024, The Ninth International Conference on Advances in Signal, Image and Video Processing // View article

Paintings-100: A Diverse Painting Dataset for Large Scale Classification

Authors:
Erica Knizhnik
Brian Rivera
Atreyee Sinha
Sugata Banerji

Keywords: painting classification, image dataset, CNN

Abstract:
Painting classification is an interesting cross-disciplinary research problem in computer vision. With the increased accessibility of digitized collections of fine-art paintings, development of effective painting classification algorithms has become vital as they have many potential applications in museums, various industries, painting theft investigation, forgery detection, art education, etc. However, the availability of large scale annotated benchmark datasets with high-resolution authentic painting images still remains a challenge. Towards that end, in this work, we develop an image dataset consisting of high-resolution painting images from 100 different artists spanning 14 different styles. This dataset is an extension of the Painting-91 dataset constructed by Khan et al. Our contribution towards extending this dataset are threefold. First, we address the limitations of the dataset by removing errors and enhancing image resolutions. Second, we add more images to augment some of the artist categories with fewer images. Third, we include the works of nine more painters from diverse backgrounds and styles for creating a more representative and inclusive database of fine-art paintings. We also perform a preliminary evaluation of this newly constructed Paintings-100 dataset using several different Convolutional Neural Network (CNN)-based classification techniques for artist recognition. Furthermore, we demonstrate that our proposed and improved dataset is more suitable for patch-based models than the earlier published Painting-91 dataset due to larger image resolutions.

Pages: 12 to 15

Copyright: Copyright (c) IARIA, 2024

Publication date: March 10, 2024

Published in: conference

ISSN: 2519-8432

ISBN: 978-1-68558-142-8

Location: Athens, Greece

Dates: from March 10, 2024 to March 14, 2024