Spot the penguin

Penguin spotting is a lonely job. In the name of science, you sit on a beach for hours at a time, watching as one penguin after another comes waddling past on their way somewhere more fun. You note down how many you saw and when, and then keep on waiting. Time passes. You chew on a piece of grass and push sand around with your toe. Finally, eventually, the penguins all walk back the other way and you note that down too. With the noting-down all taken care of you can head home, knowing not only is your job slow, it’s expensive and prone to mistakes too. Hmpgh.

Happily, it could be that biologists may not have to spend much longer getting sand in their trousers. A team from the University of Bristol have built a system that uses CCTV and computer intelligence to recognise one African penguin from another. That way they can leave the spotting to a cheaper, more accurate system, and get on with more important work themselves.

Identity parade

The way it works is by recognising patterns. You point a camera at the place where you think some penguins are going to be. The Bristol team study a group of penguins in South Africa who live near one main path down to a beach, so most of their penguins will walk down the path over the course of a day. When one shuffles past, the software hooked up to the camera checks to make sure it’s a penguin. It does this by looking for the colouring pattern that African penguins all share – black head, white neck, then a ring of black around a white tummy. Their stomachs aren’t completely white though, and that little bit of colouring opens up another trick. It lets the software recognise which penguin is which.

African penguins all have a set of spots on their stomach that’s unique to them, like fingerprints to humans. It’s a random constellation of markings made by chemistry when their coats are forming in the womb. Other animals have identifying spots too, like cheetahs, tigers and even moths. Because these markings are unique to each individual, if you can recognise the patterns you’ll know who’s who. The software analyses the bellies of each penguin it sees until it has a numerical ‘picture’ of the spots. Then it compares it to the numbers for other penguins to see whether the set of spots it’s looking at is already in its census of the penguins’ community. If it’s not then the computer stores the pattern as a new penguin. If the pattern’s familiar the computer will say who it belongs to and when they were last seen.

Penguin friends

Those unique spots that identify each penguin will help scientists learn more about their lives. Neill Campbell, a computer scientist and member of the team, explains that once the system can recognise individuals, “you can start to say not just ‘it’s a tiger or it’s a penguin’, you can start to give them names and say ‘this specific penguin has been here before’”. According to Neill that means scientists will be able to look at questions they haven’t been able to answer before. “Does a penguin have a friendship group?” he asks. “Does it go and hunt with the same group of friends each day?” Biologists could also find out whether penguins come back to the same places every year at the same time.

At the moment the best way scientists have to identify penguins is to put a tag or a band on them, but tagging tends to shorten their lifespan. So putting them on camera instead makes studying them more humane, and will help us find out more about their society. What’s more, the smart CCTV system will free some lonely biologists from penguin-watching duty. After all, no one should have to be bored at the beach.