← Portfolio
Paper

Color and emotion in pandemic art

Industry
Art and culture · research
Type of work
Computer vision paper
Period
2021 – 2022
Read PDF ↗ View on arXiv ↗

Architecture

COVID Art Museum 1,537 Instagramimages Emotions CNN trainedon ArtEmis Color clustering and Ittenpalette → harmonies Fusion KDE per emotion+ apriori rules Findings warm colors,analogous harmonies,positive emotions arXiv:2203.13770 · cs.CV · 2022

Goal

Understand the use of color and its relation to emotion in the COVID Art Museum images on Instagram. Finding: warm colors, analogous over complementary harmonies, and positive emotions prevail.

What I did

  • I'm a co-author of the paper, published on arXiv (cs.CV) in March 2022.
  • The method runs two branches in parallel: emotion classification with a CNN trained on ArtEmis, and color and harmony extraction with clustering over the Itten palette.
  • Both are then fused with per-emotion densities (KDE) and association rules (apriori).

Stack

PythonCNNArtEmisK-meansApriori