FindingEmo: An Image Dataset for Emotion Recognition in the Wild

TL;DR

FindingEmo is a comprehensive image dataset designed for emotion recognition in complex, naturalistic scenes involving multiple people, with detailed annotations on valence, arousal, and emotion labels, supporting advanced research in social and environmental contexts.

Contribution

It introduces a novel, large-scale dataset with annotations for complex scenes, expanding emotion recognition beyond faces to entire social environments.

Findings

Dataset contains 25,000 images with detailed annotations.

Includes annotations for Valence, Arousal, and Emotion labels.

Provides source code and image URLs for reproducibility.

Abstract

We introduce FindingEmo, a new image dataset containing annotations for 25k images, specifically tailored to Emotion Recognition. Contrary to existing datasets, it focuses on complex scenes depicting multiple people in various naturalistic, social settings, with images being annotated as a whole, thereby going beyond the traditional focus on faces or single individuals. Annotated dimensions include Valence, Arousal and Emotion label, with annotations gathered using Prolific. Together with the annotations, we release the list of URLs pointing to the original images, as well as all associated source code.