Clustering of illustrations by atmosphere using a combination of supervised and unsupervised learning

TL;DR

This paper proposes a hybrid supervised and unsupervised learning approach using pseudo-labels to cluster illustrations by atmosphere, improving human-like clustering performance over traditional methods.

Contribution

The paper introduces a novel combination of supervised and unsupervised learning with pseudo-labels for clustering illustrations by atmosphere, addressing label ambiguity and low-level feature limitations.

Findings

Outperforms conventional clustering methods in human-like grouping

Effective in handling ambiguous atmospheres in illustrations

Utilizes pseudo-labels to enhance feature vector quality

Abstract

The distribution of illustrations on social media, such as Twitter and Pixiv has increased with the growing popularity of animation, games, and animated movies. The "atmosphere" of illustrations plays an important role in user preferences. Classifying illustrations by atmosphere can be helpful for recommendations and searches. However, assigning clear labels to the elusive "atmosphere" and conventional supervised classification is not always practical. Furthermore, even images with similar colors, edges, and low-level features may not have similar atmospheres, making classification based on low-level features challenging. In this paper, this problem is solved using both supervised and unsupervised learning with pseudo-labels. The feature vectors are obtained using the supervised method with pseudo-labels that contribute to an ambiguous atmosphere. Further, clustering is performed based on these feature vectors. Experimental analyses show that our method outperforms conventional methods in human-like clustering on datasets manually classified by humans.