Classifying Dog Emotions from Facial Expressions: Automated Facial Landmark Analysis vs. Manual Coding
Array · 2026-11 · 101482
DOI: 10.1016/j.array.2026.101482 ↗At a glance
Journal article in Array, 2026-11. Written by Jiao-Ling Appels and 3 co-authors. 74 references. The full text is freely available (open access, CC BY 4.0).
Abstract (original language)
Emotional expression in dogs is central to dog-human interactions. Reliable indicators are essential for interpreting animal emotions; however, their predictive value remains underexplored. Fireworks, which elicit fear in many dogs, provide a suitable real-world context for examining affective states. We used machine learning to classify firework vs. non-firework situations from dogs’ behaviour and expressions using two approaches: (1) automated facial landmark detection and (2) ethogram-based manual coding of only facial behaviours. The Random Forest model, based on manual coding, achieved high cross-validated generalisation performance using subject-wise hold-out validation (accuracy = 0.90 and predictive validity = 0.80), identifying different ear positions and blinking as key fear indicators. The best generalised automated facial landmark model using the same validation achieved 0.70 accuracy and 0.74 predictive validity. This exploratory comparison suggests that manual coding achieved superior classification performance, perhaps because it captures events such as blinking that are too brief for significantly altered landmark configurations, or due to challenges posed by the high morphological diversity of the subjects. Nonetheless, our findings indicate that both automated facial landmark analysis and manually coded facial expressions can distinguish between firework and control conditions with high accuracy, offering potential for future automated welfare monitoring.
Cite
APA
Appels, J., Martvel, G., Zamansky, A., & Riemer, S. (2026). Classifying Dog Emotions from Facial Expressions: Automated Facial Landmark Analysis vs. Manual Coding. Array, 101482. https://doi.org/10.1016/j.array.2026.101482
Vancouver
Appels J, Martvel G, Zamansky A, Riemer S. Classifying Dog Emotions from Facial Expressions: Automated Facial Landmark Analysis vs. Manual Coding. Array. 2026:101482. doi:10.1016/j.array.2026.101482
BibTeX
@article{Appels2026,
author = {Appels, Jiao-Ling and Martvel, George and Zamansky, Anna and Riemer, Stefanie},
title = {Classifying Dog Emotions from Facial Expressions: Automated Facial Landmark Analysis vs. Manual Coding},
journal = {Array},
year = {2026},
pages = {101482},
doi = {10.1016/j.array.2026.101482}
}
EUVIMED