Using machine learning to distinguish between authentic and imitation Jackson Pollock poured paintings: Art images
Data files
Jul 01, 2024 version files 42.53 GB
Abstract
Jackson Pollock’s abstract poured paintings are celebrated for their striking aesthetic qualities. They are also among the most financially valued and imitated artworks, making them vulnerable to high-profile controversies involving Pollock-like paintings of unknown origin. Given the increased employment of artificial intelligence applications across society, we investigate whether established machine learning techniques can be adopted by the art world to help detect imitation Pollocks. The low number of images compared to typical artificial intelligence projects presents a potential limitation for art-related applications. To address this limitation, we develop a machine learning strategy involving a novel image ingestion method which decomposes the images into sets of multi-scaled tiles. Leveraging the power of transfer learning, this approach distinguishes between authentic and imitation poured artworks with an accuracy of 98.9%. The machine also uses the multi-scaled tiles to generate novel visual aids and interpretational parameters which together facilitate comparisons between the machine’s results and traditional investigations of Pollock’s artistic style.
README: Art Images
https://doi.org/10.5061/dryad.m905qfv91
This data contains all the individual tiles of the art images used for the paper "Using Machine Learning to Distinguish Between Authentic and Imitation Jackson Pollock Poured Paintings: A Tile-Driven Approach to Computer Vision"
Description of the data and file structure
Each art image is cropped and then tiled at multiple physical size scales. Each zipped folder contains all the tiles for all the images at a particular size scale (e.g. folder "20" refers to a 20cm x 20cm square tile). The range of tile sizes is from 10cm to 360cm every 5cm and "Max".
Due to the dryad storage limitations the "10" folder was split into several zipped folders, "10_ACDEF", "10_G", "10_J", and "10_P" .
They should be combined into one single folder labeled "10".
The final folder structure should be as follows
"ImageClassifier/Paintings/Processed/Raw/ (all tile size folders)"
When running code none of the subfolders in "Raw" should be zipped
In "ImageClassifier" should exist all the code from github (linked in this repo)
All files that start with a "P" or a "J" are known Pollock art works.
Those that start with other letters are known to be done by other artists
"Paintings512.zip" should be extracted in the parent "ImageClassifier" directory
Sharing/Access information
The images of the 588 artworks used in our study were acquired in collaboration with The Pollock-Krasner Foundation, The Pollock-Krasner Study Center, The International Foundation for Art Research, and Francis V. O’Connor (chief Pollock connoisseur and co-author of the Catalogue Raissonne). The collection and analysis method of all images complies with the terms and conditions for the sources of the data.
The S1 Table in the associated manuscript provides a comprehensive list of the image sets. The image sets feature 2 overall categories of artwork - those established as being created by Pollock and those established to be by other artists.
Code/Software
All Code can be found in the linked github repository
Methods
The images of the 588 artworks used in our study were acquired in collaboration with The Pollock-Krasner Foundation, The Pollock-Krasner Study Center, The International Foundation for Art Research, and Francis V. O’Connor (chief Pollock connoisseur and co-author of the Catalogue Raissonne). The collection and analysis method of all images complies with the terms and conditions for the sources of the data.
The S1 Table provides a comprehensive list of the image sets. The image sets feature 2 overall categories of artwork - those established as being created by Pollock and those established to be by other artists.