An automated facial recognition coding system uses images and videos to determine specific emotions in livestock
Advances in precision agriculture have made it possible to track beef or dairy cow movements, feed and water consumption, body weight and basic behaviour patterns, which can help limit disease and increase production.
Radio-frequency identification (RFID) tags and readers, electronic scales and implanted chips send data to smartphones and computers when locations or routine behaviours vary outside the norm.
But livestock technology isn’t stopping there.
The 1990s brought visual and pattern recognition combined with digital photography to help identify individual animals by their unique physical characteristics. This ability helped track movement and was a step in providing a higher quality of welfare.
“While these older methods of data collection sacrificed a small amount of animal welfare for the benefit of the population at large, no such compromises need be made in the future,” says Dr. Suresh Neethirajan, associate professor at the Wageningen University in the Netherlands.
Neethirajan is developing Wur Wolf, an automated facial recognition coding system using images and videos to determine specific emotions. The platform uses sensors and harnesses the potential of machine learning to make automated monitoring a reality.
Facial recognition has long been used in day-to-day human context for various applications including phone password protection, law enforcement, detection of genetic disorder phenotypes, disease diagnosis and measuring tourist enjoyment levels and shopper satisfaction.
With domestic animal care, farm workers typically use a combination of experience, observation, basic measurements and gut feelings to monitor health and welfare. To avoid unnecessary handling in obtaining functional data and to reduce the human element of subjectivity with manual assessments, these technologies can work together to identify responses such as distress or pain.
“One of the great benefits of this emerging technology for animal monitoring is the potential for non-invasive identification. Not only is this humane, but it also has distinct financial benefit as happier animals may be more productive in general. Emotion measurement is one concrete step farmers can take toward improving care.”
Neethirajan also believes emotion identification can improve animal-human interactions. Precision livestock, artificial intelligence and digital technologies will transform domestic animal care, he says.
Stress and pain are the largest contributors to illness and disease in livestock. Domestic animals used for food also yield higher quality products when content with their surroundings and situation. Early signs of stress provide quicker opportunities to address stress or illness. Detection in individuals using facial identification can be used to categorize pain and discomfort and alert workers of impending dystocia.
Wur Wolf, developed by the Farmworx group at the Wageningen University, evaluates 14 facial feature combinations and seven emotional states using a database of images and videos of thousands of cattle and hogs.
Understanding and interpreting the emotional standing of domestic animals is much more difficult than in humans because people portray a wide array of expressions communicating emotion and intent. Neethirajan says each animal has a unique way of showing facial expression.
“To ensure access to sustainable and high-quality health attention and welfare in animal husbandry management, innovative tools are needed. Unlocking the full potential of automated measurement of mental and emotional states through digitalization such as facial coding systems will help to blur the lines between biological, physical, and digital technologies.”
The expanding Wur Wolf database uses four principal facial expressions: neutral, aggression, happiness and fear. Its non-invasive method analyzes and compares facial features and movements. A computerized determination of an animal’s current state is then made using what is referred to as “the grimace scale.” The software uses a series of points in relation to phenotypic features and then bases emotional states on their location.
The grimace scale analyzes physical reactions associated with varying levels of discomfort. They are species-specific, normally focusing on tension in the neck, shape of the eye, strain in the brow, nose bunching, and positioning of the ears. These visual cues combined with vocal signals help identify discomfort levels.
Scale measurements have been accepted as an accurate way to quantify strong responses. They can be analyzed with recognition software and assisting algorithms to analyze and interpret facial structures.
The program also reduces the need for animal and human interaction that changes behaviour and adds stress. Emotional status is determined from a safe distance. Animals within large groups can also be identified individually and observed.
Neethirajan says humans naturally display empathy toward other non-human species, thus human error bias tends to occur when observing non-human behaviour. Facial recognition helps reduce this unintentional bias and limits improper interpretation.
“With well-calculated software, these discrepancies will cease to exist, and researchers will focus more of their time on finding significant points and connections within recorded data, rather than spending their time actually recording the data,” he adds.
“Animals can’t tell humans how much pain they’re in, so it’s up to us to interpret the level they’re experiencing and treat it appropriately. This task is most accurately completed when emotions are clearly and quickly detectable.”
Neethirajan says systems such as WUR Wolf represent the future in technological solutions.
“Non-invasive technology to assess good and poor welfare of farm animals, including positive and negative states, will soon be possible using this platform. The ability to track and analyze how they feel will be a breakthrough in establishing animal welfare auditing tools. New sensing technologies will likely play a massive role in the digital future of agriculture, including identifying and rating emotions. Artificial intelligence and sensor-based facial recognition will become an every day tool predicting livestock behaviour through their precise measurements.”