Scientists programmed a computer to recognize sheep facial expressions, based on techniques for human face recognition. “Our multi-level approach starts with detection of sheep faces, localisation of facial landmarks, normalisation and then extraction of facial features,” says the study. Each of these items is associated with a different level of sheep pain, according to SPFES.
The computer system was able to classify 9 different sheep faces and consequently estimate their pain levels. The overall accuracy was 67 percent, which was calculated by dividing the number of detections the computer got right (double checked by a human) by the total sample size.
It’s not a breathtaking result, but it certainly is an encouraging one. The team believes they can increase accuracy with more data, and that this technique could be extended to apply to other animals such as mice, rabbits and horses. It’s a preliminary result, to be sure, but eventually this could make it much easier for veterinarians to diagnose and treat animals.