Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Deep Ensemble Art Style Recognition

Orfeas Menis-Mastromichalakis, Natasa Sofou, Giorgos Stamou

TL;DR

This paper tackles automated art style recognition by evaluating eight CNN architectures with transfer learning across Pandora 18K and WikiArt datasets. It introduces a stacking ensemble that fuses diverse model outputs via a shallow meta-classifier, achieving state-of-the-art performance on WikiArt (68.55%) and strong results on Pandora 18K (72.47%). The study also analyzes data preprocessing effects (resize vs crop) and demonstrates how multiple perspectives on style can be complementary. Overall, the work advances robust, data-driven art style classification and suggests practical paths for deploying ensemble classifiers alongside human experts in art analysis.

Abstract

The massive digitization of artworks during the last decades created the need for categorization, analysis, and management of huge amounts of data related to abstract concepts, highlighting a challenging problem in the field of computer science. The rapid progress of artificial intelligence and neural networks has provided tools and technologies that seem worthy of the challenge. Recognition of various art features in artworks has gained attention in the deep learning society. In this paper, we are concerned with the problem of art style recognition using deep networks. We compare the performance of 8 different deep architectures (VGG16, VGG19, ResNet50, ResNet152, Inception-V3, DenseNet121, DenseNet201 and Inception-ResNet-V2), on two different art datasets, including 3 architectures that have never been used on this task before, leading to state-of-the-art performance. We study the effect of data preprocessing prior to applying a deep learning model. We introduce a stacking ensemble method combining the results of first-stage classifiers through a meta-classifier, with the innovation of a versatile approach based on multiple models that extract and recognize different characteristics of the input, creating a more consistent model compared to existing works and achieving state-of-the-art accuracy on the largest art dataset available (WikiArt - 68,55%). We also discuss the impact of the data and art styles themselves on the performance of our models forming a manifold perspective on the problem.

Deep Ensemble Art Style Recognition

TL;DR

This paper tackles automated art style recognition by evaluating eight CNN architectures with transfer learning across Pandora 18K and WikiArt datasets. It introduces a stacking ensemble that fuses diverse model outputs via a shallow meta-classifier, achieving state-of-the-art performance on WikiArt (68.55%) and strong results on Pandora 18K (72.47%). The study also analyzes data preprocessing effects (resize vs crop) and demonstrates how multiple perspectives on style can be complementary. Overall, the work advances robust, data-driven art style classification and suggests practical paths for deploying ensemble classifiers alongside human experts in art analysis.

Abstract

The massive digitization of artworks during the last decades created the need for categorization, analysis, and management of huge amounts of data related to abstract concepts, highlighting a challenging problem in the field of computer science. The rapid progress of artificial intelligence and neural networks has provided tools and technologies that seem worthy of the challenge. Recognition of various art features in artworks has gained attention in the deep learning society. In this paper, we are concerned with the problem of art style recognition using deep networks. We compare the performance of 8 different deep architectures (VGG16, VGG19, ResNet50, ResNet152, Inception-V3, DenseNet121, DenseNet201 and Inception-ResNet-V2), on two different art datasets, including 3 architectures that have never been used on this task before, leading to state-of-the-art performance. We study the effect of data preprocessing prior to applying a deep learning model. We introduce a stacking ensemble method combining the results of first-stage classifiers through a meta-classifier, with the innovation of a versatile approach based on multiple models that extract and recognize different characteristics of the input, creating a more consistent model compared to existing works and achieving state-of-the-art accuracy on the largest art dataset available (WikiArt - 68,55%). We also discuss the impact of the data and art styles themselves on the performance of our models forming a manifold perspective on the problem.
Paper Structure (18 sections, 6 figures, 7 tables)

This paper contains 18 sections, 6 figures, 7 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Datasets
  4. Dataset 1
  5. Dataset 2
  6. Method
  7. Baseline
  8. Simple Ensemble
  9. Stacking Ensemble
  10. Experimental Results
  11. Setting the baseline
  12. Data Preprocessing
  13. Architectures
  14. Proposed Methodology
  15. Discussion
  16. ...and 3 more sections

Figures (6)

  • Figure 1: Similar paintings-Different styles
  • Figure 2: Number of artworks per class in Dataset 1.
  • Figure 3: Number of artworks per class in Dataset 2.
  • Figure 4: The proposed stacking ensemble method.
  • Figure 5: Data augmentation levels on a painting
  • ...and 1 more figures