AI interpreting animal emotions

A recent profile on in this area, featured in Science, highlights how researchers are using AI to assess animal emotions more accurately and efficiently than humans. One such system under development is Intellipig, an AI-driven monitoring platform designed to capture and analyse health and welfare data in pigs automatically.

Why is pig health monitoring important?

In 2018, the University of the West of England proposed a research project focused on early detection of pig health issues. The goal was to improve animal well-being and reduce production costs using machine vision – a low-cost, non-intrusive technology.

Monitoring pig welfare is crucial for farmers, as it enables them to track herd health, growth, and productivity. With accurate data, farmers can make informed decisions to improve animal welfare, optimise operations, and enhance profitability. Identifying trends and potential issues early can lead to better breeding programmes, disease prevention, and efficient feed usage.

Continuous monitoring of farm animal health and welfare depends on practical, reliable measurement tools. While animal behaviour is occasionally recorded, it is rare for data to provide insights into an individual animal’s emotional state. Intellipig aims to change this by integrating AI-based observation techniques with real-time monitoring.

How Intellipig works

Intellipig employs facial recognition and machine learning to assess both an animal’s emotional state and body condition. By capturing high-resolution images and analysing subtle facial expressions, the system can detect signs of stress, discomfort, or illness. Researchers have successfully developed algorithms that identify individual pigs based on facial biometrics, detect changes in their expressions, and assess body condition scores and weight estimation.

To enhance monitoring efficiency, Intellipig integrates additional data points, such as food and water intake, movement patterns, and growth rates. AI-driven predictive analytics can identify deviations from normal behaviour, allowing for early detection of potential health issues. The system is designed to operate autonomously, collecting data continuously and alerting farmers when intervention may be necessary. This data-driven approach enables proactive health management, reduces the reliance on manual observation, and contributes to overall farm efficiency.

AI in recognising animal discomfort

Beyond farm animals, AI is also being explored for use with pets. At the University of Haifa, researchers are developing AI-based recognition software to identify signs of discomfort in dogs’ facial expressions. Their system utilises computer vision techniques to analyse subtle changes in facial tension, eye shape, and ear positioning, which may indicate pain or distress.

The system is being refined to ensure accuracy across different breeds and lighting conditions, and by training deep learning models on vast datasets of canine expressions, researchers aim to create a tool that can assist veterinarians and pet owners in identifying health concerns earlier.

However, researchers caution that AI models must be carefully trained to avoid unintended biases. For instance, if images of distressed animals are predominantly taken at night while happy ones are captured during the day, the algorithm might misinterpret lighting conditions as an emotional indicator. Addressing such challenges requires diverse datasets and continual refinement of deep learning techniques.

AI’s role in interpreting animal emotions is still evolving, but these advancements suggest a future where technology could help humans better understand and care for animals, whether on farms or in homes. As AI models become more sophisticated, their ability to detect and respond to animal emotions could lead to improved welfare standards and more humane treatment across various settings.