At the intersection of technology and animal behavior science, Albert C. Villaluz and colleagues presented groundbreaking research at the 2024 9th International Conference on Intelligent Information Technology. Their study applied machine learning and computer vision to decode emotional states in domestic dogs based on visible body and facial cues.
The team designed a multi-stage process that included data collection from videos and images, followed by detailed annotation, preprocessing, and augmentation to ensure a diverse and representative dataset. They utilized YOLOv5 for object detection to identify and track dogs within frames and implemented transfer learning using the MobileNet architecture to classify four emotional categories: aggressiveness, fear, anxiety, and happiness.
Performance evaluation revealed that including facial feature data substantially improved model accuracy compared to body-only analysis. This highlights the importance of integrating multi-modal data—combining facial and postural information—to better capture the subtle cues that reflect a dog’s affective state.
While the study primarily focused on happiness as a test case, it sets the foundation for future systems that can non-invasively assess a dog’s emotional state in real-time. Such advancements could revolutionize welfare monitoring, behavioral research, and training practices by providing objective indicators of emotional well-being.
Villaluz et al. emphasize that these technological innovations aim not to replace human intuition but to enhance empathy and understanding between humans and their canine companions through scientifically grounded tools. By bridging artificial intelligence with ethology, this research demonstrates the expanding role of technology in promoting humane and evidence-based animal care.
Source: Villaluz, A. C., Goma, J. C., Besa, J. V. T., Ignacio, J. I. D., & Zaguirre, S. A. A. (2024). Emotion Classification in Domestic Dogs Using Computer Vision Based on the Dog’s Body and Face. In Proceedings of the 2024 9th International Conference on Intelligent Information Technology. Published February 23, 2024.
