Automated Facial Landmark Analysis vs. Manual Coding: Accuracy in Dog Emotional Expression Classification

Published: 23 October 2025| Version 1 | DOI: 10.17632/swk97wxzzg.1
Contributors:
, George Martvel, Anna Zamansky,

Description

This dataset contains behavioural and facial landmark data from dogs recorded during a firework situation and a control evening without fireworks. The Excel sheet includes manually coded behaviours based on a predefined ethogram, along with descriptions of all variables used in the analysis. Additionally, two ZIP folders contain CSV files with facial landmark coordinates extracted from video frames. These datasets were analysed using two Python scripts, each corresponding to one analytical approach: (1) manual behaviour coding and (2) automated facial landmark analysis.

Files

Institutions

  • Veterinarmedizinische Universitat Wien
  • University of Haifa
  • Jonkoping University

Categories

Animal Welfare, Behavior Coding, Applied Machine Learning

Licence