Computer Vision and Visual Arts: Automated Analysis of Artistic Style, Color, and Composition
Qubahan Journal of Social Science · 2025
DOI: 10.48161/qjss.v2n4a10 ↗Kurz gefasst
Fachartikel in der Zeitschrift Qubahan Journal of Social Science, 2025. Verfasst von Maryna Shlenova. Die Arbeit nennt 32 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC-ND 4.0).
Abstract (Originalsprache)
This paper presents an empirical computer-vision framework for automatically analyzing visual artworks using style, color, texture, composition, and visual-similarity indicators. The research addresses the challenge of examining large digitized art collections at scale while preserving art-historical interpretability. The empirical analysis uses a balanced public artwork-image corpus of 60,000 images, evenly distributed across 10 artistic style categories (6,000 images per class). The official dataset partition is split into a 40,000-image training set, a 10,000-image validation set, and a 10,000-image testing set. The framework extracts 94 interpretable visual descriptors spanning color histograms, entropy, edge density, local variation, symmetry, quadrant balance, and central contrast. Among the evaluated classifiers, Extra Trees performed best, with a test accuracy of 32.75% and a macro F1-score of 30.84%, significantly exceeding the 10.0% random and majority-class baselines. Feature-group ablation and importance analyses show that color descriptors contribute most, accounting for 76.78% of total feature importance. The study provides a reproducible empirical workflow linking measurable image features, statistical validation, and art-historical interpretation for scalable cultural heritage analysis.
Zitieren
APA
Shlenova, M. (2025). Computer Vision and Visual Arts: Automated Analysis of Artistic Style, Color, and Composition. Qubahan Journal of Social Science. https://doi.org/10.48161/qjss.v2n4a10
Vancouver
Shlenova M. Computer Vision and Visual Arts: Automated Analysis of Artistic Style, Color, and Composition. Qubahan Journal of Social Science. 2025. doi:10.48161/qjss.v2n4a10
BibTeX
@article{Shlenova2025,
author = {Shlenova, Maryna},
title = {Computer Vision and Visual Arts: Automated Analysis of Artistic Style, Color, and Composition},
journal = {Qubahan Journal of Social Science},
year = {2025},
doi = {10.48161/qjss.v2n4a10}
}
EUVIMED