A research team from the Federal Polytechnic School of Lausanne (EPFL) has developed technology based on artificial intelligence (AI) to better decipher the body language of animals. This model would be ten to 100 times more efficient than current tools.
THE Mackenzie Mathis’ laboratory has developed a new “open source” tool called SuperAnimal. This method is capable of automatically recognizing, without human supervision, the location of key points, notably joints, on more than 45 animal species, EPFL explained in a press release on Friday.
External content
This external content cannot be displayed because it may collect personal data. To view this content you must authorize the category Social networks.
Accept More information
It is possible today to create deep learning models, but they rely on intense human work intended to identify key points on each animal. This can lead to confusion, especially when combining databases. SuperAnimal “standardizes all of this and makes our model 10 to 100 times more efficient than current tools,” writes Mackenzie Mathis, whose works are published in Nature Communications.
Health
Researchers point out that behavioral analysis can provide a lot of information on the state of health or the intentions of a living being. It’s difficult to get a horse to tell if it’s happy or if it’s in pain by putting its leg down. But his body will express the response in its own way, through its movements.
For those who know how to recognize them, a limp or facial expressions can provide interesting elements: “Artificial intelligence has the capacity to automate the analysis of these movements, however subtle they may be,” notes the press release. Such tools will also help remove observational bias and help humans draw faster and more reliable conclusions.
>> Read also: Science is evolving about animal consciousness
Expected developments
The SuperAnimal method is an evolution of a technique that Mackenzie Mathis’ laboratory had already distributed under the name “DeepLabCut“These advances, accessible to the scientific community, should be of particular interest to veterinarians, as well as biomedical research, when it comes to observing the behavior of laboratory mice, observes EPFL.
Other species, such as birds, fish or insects, should soon be integrated into the model.
Better understanding what animals say with their bodies could be useful for their protection, biomedicine and neuroscience research.
sjaq and ats
