Artwork Categorization
Abdullah Ahmed, Alec Albrecht, Carlos Hernandez, Ankita Somu, Sanjana Srinivasan
Project Proposal
Introduction/Background
With the rise of artwork generated by text-to-image models like DALL-E and Midjourney, it raises an interesting question about influences from source material. Generative art models are trained on large datasets including works from different art movements over time. Our team seeks to develop a solution for classifying an art piece into a stylistic period. This categorization algorithm could be used to identify and retrace the art styles mimicked in generative art. Previous approaches have used k-Nearest Neighbor and support vector machine algorithms to build their models (Falomir, Zoe, et al.).
Problem definition
Given an image our project aims to classify the style of an art piece. Our goal is to create a model that consistently classifies the style of the images in any collection.
Methods
We identified a dataset on Kaggle that collects artwork for the 50 most (subjectively) influential artists, along with their respective, time period, style, and other classifiers. Based on this choice of dataset, many of these classifiers can already be used as clusters, such as 50 artist clusters. As a result, we can apply different unsupervised learning techniques to the full data set with a known number of clusters to see if different clustering algorithms are able to correctly group images by the classifier. Scikit Learn provides many implementations for these algorithms. Additionally, a very common approach to supervised learning with images is a Convolution Neural Network. Support vector machines are an especially powerful algorithm for classification and may be better suited toward solving our problem. We will choose to use one of these models based on future research and/or timeline constraints. Going forward for both supervised and unsupervised approaches, it may be necessary to simplify the problem to a smaller group of artists or styles first (such as differentiating between 2 of them) and then expand outward given the lessons learned from the simplified model.
Potential results and discussion
The goal for this project is to create a model that can accurately determine the style of a piece of art in any given collection. As museums and galleries transition more into digitized collections, the identification of art style can also assist in organizing and curating pieces, as well as categorizing newly collected artworks based on style, which is something even humans have difficulty with. This model could be critical in Visual Arts industries as it can aid not only in labeling art pieces, but also in identifying forgeries.
The consistency measure our model is expected to hit is anywhere between 85-97%, meaning that our predictions of the average or squares will only deviate from the true values 3-15% of the time. Similarly, we also hope our precision score is as close to 1 as possible, where we will consider success to be anything greater than or equal to 0.80.
Though complex, it would be intriguing to investigate the possibility of extending our model to identifying famous artists of paintings based on art styles as well. This could then be utilized to see what nuances may distinguish authentic paintings from supposed replicas.
Proposed timeline
Contribution table
| Name | Contribution |
|---|---|
| Abdullah Ahmed | Tech Lead |
| Alec Albrecht | Methods and Algorithms |
| Carlos Hernandez | Project Manager |
| Ankita Somu | Potential Results |
| Sanjana Srinivasan | Project Analysis & Discussion |
References
Categorizing Paintings in Art Styles Based on Qualitative Color Descriptors, Quantitative Global Features and Machine Learning (QArt-Learn)
Falomir, Zoe, et al. “Categorizing Paintings in Art Styles Based on Qualitative Color Descriptors, Quantitative Global Features and Machine Learning (QArt-Learn).” Expert Systems with Applications, vol. 97, 2018, pp. 83–94, https://doi.org/10.1016/j.eswa.2017.11.056.
Using Machine Learning for Identification of Art Paintings
Blessing, Alexander. “Using Machine Learning for Identification of Art Paintings.” (2010).
Discerning Art Works through Active Machine Learning
Z. Yu, “Discerning Art Works through Active Machine Learning,” 2022 3rd International Conference on Computer Vision, Image and Deep Learning & International Conference on Computer Engineering and Applications (CVIDL & ICCEA), Changchun, China, 2022, pp. 1002-1006, doi: 10.1109/CVIDLICCEA56201.2022.9824180.
Best Artworks of All Time
Icaro (2019, February). Best Artworks of All Time, Version 1. Retrieved October 6, 2023 from https://www.kaggle.com/datasets/ikarus777/best-artworks-of-all-time.