Reproduce “CREMD: Crowd-Sourced Emotional Multimodal Dogs Dataset” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Jinho Baek, Houwei Cao, Kate Blackwell

Researchers from New York Institute of Technology developed CREMD, a crowd-sourced multimodal dataset, to analyze how presentation modalities and annotator demographics influence the perception and labeling of dog emotions. The study revealed that visual context significantly improved annotation agreement, and professional annotators demonstrated the highest consistency in identifying canine emotional states.

Key insights

  • Adding visual context to dog videos significantly improved inter-rater agreement for emotion annotations, though the isolated impact of audio on agreement was inconclusive.
  • Annotator demographics strongly influenced agreement: non-dog owners and male annotators exhibited higher consistency than dog owners and female annotators.
  • Professional experience in animal-related fields consistently led to the highest inter-rater agreement, suggesting a more objective and consistent understanding of canine behavior.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.