Compare the performance of the models in art classification.

Area of Science:

• Computer Vision
• Digital Humanities
• Art History

Background:

• Museums house vast art collections requiring classification.
• Digitization of artworks is increasing, necessitating automated analysis tools.
• Computer-assisted art classification aids researchers and the public.

Purpose of the Study:

• To compare the performance of 7 different models for art classification.
• To evaluate the impact of transfer learning on art classification accuracy.
• To achieve state-of-the-art performance in classifying art genres, styles, and artists.

Main Methods:

• Tested 7 distinct computational models across 3 diverse art datasets.
• Implemented and compared models with and without transfer learning.
• Visualized classification processes to understand model behavior and limitations.

Main Results:

• Optimized model structures significantly improved classification performance.
• Achieved state-of-the-art results across all genre, style, and artist classification tasks.
• Transfer learning enhanced model capabilities for art analysis.

Conclusions:

• Computational models can effectively classify digitized artworks by genre, style, and artist.
• Model optimization and transfer learning are key to advancing automated art analysis.
• Visualizations provide insights into the challenges of computational art classification.