If you have a cat, you may often wonder what it is thinking, whether it is happy, if it is in pain, or if it wants to steal your sandwich. However, cats' expressions are often difficult to decipher, and their emotions are not easily understood by humans. Fortunately, scientists are using AI to unravel cats' minds and maybe even make them "talk".
Recently, a study used two different machine Xi algorithms to analyze the ** of 84 cats to determine if they were in pain. The cats were all sent to a veterinary hospital in Germany for various illnesses or injuries. The researchers gave each cat a pain score based on a criterion called the Glasgow Cat Comprehensive Pain Assessment Scale, as well as their condition. They then fed these score-labeled ** into two machine-Xi algorithms.
One of the algorithms is based on feature points, which will detect more than forty specific locations on the cat's face, such as eyes, nose, mouth, etc., and then judge the cat's pain level based on the changes in these positions. The other algorithm is based on deep Xi, which analyzes the whole sheet directly without extracting feature points. The results show that the algorithm based on feature points can achieve an accuracy rate of 77%, while the algorithm based on depth Xi has an accuracy rate of 65%. The researchers believe that if the deep learning Xi algorithm is trained with more **, its performance can be improved.
An interesting finding of this study is that cats' eyes are not a window to their emotions. If you want to accurately identify a cat's pain, you need to look at its mouth. The shape and open of a cat's mouth can reflect its pain level. This is a useful tip for cat owners and veterinarians.
Of course, pain is just one emotion for cats, they have many other emotions such as happiness, sadness, curiosity, anger, etc. Is there a way for AI to recognize more of the cat's emotions?The answer is possible. A researcher who was not involved in the AI study recently published a paper showing that cats are capable of producing nearly 300 different facial expressions. He hopes to work with AI researchers to create a model that can detect specific emotions in cats based on their facial expressions. In this way, we can know what a cat is feeling by looking at it.
Here's the catch, though. We know that human emotion recognition AI is very flawed because humans often disguise or disguise their emotions, such as smiling not necessarily representing happiness, but may be polite or embarrassed. Would cats be like that?Do they lie or deceive us too?This is a question to be explored. Even if it's an imperfect tool, though, emotion recognition AI is valuable because it can help us overcome communication barriers with pets.
The researchers also hope to expand their work to the study of emotions in other animals. So, in the near future, we may only need some datasets to make our pets almost "talk". Isn't that cool?Do you want to know what your cat is thinking?Feel free to leave a message in the comment section to share your thoughts.