Dr Tilo Burghardt is a Lecturer in the Department of Computer Science. He kindly took me through the story of how his PhD research (which included several months on an island in Africa looking at penguins – an unlikely task for a computer scientist) has led to his involvement in the emerging field of animal biometrics.
Imagine you’re on holiday; you take a photo of the local wildlife and, based on the image, your computer tells you not only what species the animal is but also the characteristics of this particular individual. This technology is just one of the many applications from Tilo’s work on computer vision – how machines make sense of things such as images and videos. Tilo’s research started by looking at how computers can identify species and even individual animals based on pattern recognition. But the challenge is how a technique that works on lab footage in a controlled environment can be adapted to animals in the wild. Here, machine learning – systems that learn from data to improve how they work – can be used to extract meaning from the gigabytes (and more) of data from video footage of animals. And from this meaning we can start to understand what animals are like physically and how they behave, both as individuals and as a species. This not only has great applications for tourism and entertainment; it also starts to unpick the big question of nature versus nurture and to what extent an animal's characteristics and behaviours are determined by their genetics and environment.
