Animal behaviour researchers frequently rely on hours upon hours of video material to get information on the habits of animals. They have to manually evaluate this task. Typically, this necessitates researchers laboriously jotting down observations on the animals' behaviour over the course of several weeks or months of recordings.
Now, a research team has developed an automated method for analysing these types of recordings. The team has used Computer vision and machine learning to accomplish the image-analysis method they developed. The AI algorithm could greatly contribute to AIaaS (Artificial Intelligence as a Service) Market as it can recognise individual animals and recognise certain behaviours. These indicate curiosity, fear, or harmonious social interactions with other species members.
The method developed is one-of-its-kind in terms of analyses of animal behaviour. This opens the door to longer-term, in-depth behavioural science studies and helps to enhance animal welfare. The system essentially provides scientists with a one-click option for automatically assessing video material, regardless of how long or detailed it is. Another benefit of the new method is that it is repeatable. When multiple groups of researchers use the same algorithm to evaluate their video data, comparing results is easy. This is because everything is based on the same standards.
Furthermore, because the new algorithm is so sensitive, it can detect even modest behavioural changes that occur over a period of time. These changes are often difficult to identify with the naked eye. This technology can also be used to improve animal husbandry by allowing round-the-clock surveillance to detect undesirable behaviours. Keepers can quickly intervene to improve conditions for the animals in their care by spotting negative social interactions or the onset of sickness early on.
Thus, inspiring them to improve animal care and undertake automated behavioural studies. For instance, Zoo researchers had to manually annotate nighttime video footage in a recently released study analysing patterns of elephant sleep behaviour. They believe that by using the new method, they will be able to automate and scale up similar findings in the future. Researchers' approach can detect even mild or uncommon behavioural changes in research animals, such as stress, worry, or pain indicators. This can assist in improving the quality of animal studies while simultaneously scaling down the animal used and the burden placed on them.