Recently, researchers from the University of Oxford have developed new artificial intelligence (AI) software designed to recognise and track the faces of specific individual primates such as chimpanzees in the wild. The software is thought to help researchers and wildlife conservationists alike in helping to significantly cut back on time and resources spent analysing video footage designed to track animals in their natural habitat. In turn, this will allow them to study the complex behaviours of chimpanzees and other primates in a more efficient manner.
The algorithm which was developed for this facial recognition AI, was run on 50 hours of archival footage – spanning 14 years – of chimpanzees based in Bossou in Guinea. The footage of 23 chimpanzees yielded around 10 million facial images. The developed algorithm learned to continuously track and recognise individuals from raw video footage. It is said to have performed well, even when faced with low light and poor-quality images, and successfully recognised the chimpanzees’ faces even when they weren’t necessarily looking towards the camera. Overall, the AI had an identity recognition accuracy rate of 92 per cent, correctly identifying an animal’s sex 96 per cent of the time.
The team of researchers then compared the AI technology’s ability with that of humans. By selecting 100 random still images, the AI, along with a group of people, were tasked with identifying the chimpanzee in each image. The algorithm had an 84 per cent accuracy, taking a mere 30 seconds to complete the task. On the other hand, the group of people, who were researchers specifically experienced in recognising the chimps, spent 55 minutes completing the task, and achieved a meagre 42 per cent accuracy rate on average.
Applicants for Computer Science, along with students applying for Biological Sciences, can delve deeper into exploration on this topic, considering how the development of such AI technologies could facilitate improved efficiency in a variety of research fields such as animal behaviour analysis, particularly through the use of data collection and the rise of ‘big data’.