Deep learning helps survey endangered animals from space
Image credit: Izanbar/Dreamstime
Scientists have successfully used satellite images coupled with deep learning to count animals in complex geographical landscapes. This system could help conservationists monitor populations of endangered species.
As part of this research, the satellite Worldview-3 used high-resolution imagery to capture African elephants moving through forests and grasslands. The researchers said the automated system detected animals with the same accuracy as humans are able to achieve.
The project was conducted by the University of Bath in collaboration with the University of Oxford and the University of Twente in the Netherlands. Dr Olga Isupova, a computer scientist at the University of Bath, created the algorithm that enabled the detection process.
Dr Isupova said the new surveying technique allows vast areas of land to be scanned in a matter of minutes, offering a much-needed alternative to human observers counting individual animals from low-flying aircraft. As it sweeps across the land, a satellite can collect over 5,000 km² of imagery every few minutes, eliminating the risk of double counting. Where necessary (for instance, when there is cloud coverage), the process can be repeated the next day, on the satellite’s next orbit of Earth.
The population of African elephants has nose-dived over the past century – this is mainly due to poaching and habitat fragmentation. With only 40,000-50,000 elephants left in the wild, the species is classified as endangered. “Accurate monitoring is essential if we’re to save the species,” said Dr Isupova. “We need to know where the animals are and how many there are.”
Satellite monitoring eliminates the risk of disturbing animals during data collection and ensures humans are not hurt in the counting process, according to the researchers. It also makes it simpler to count animals moving from country to country, as satellites can orbit the planet without regard for border controls or conflict.
Although the study was not the first to use satellite imagery and algorithms to monitor species, it was the first to reliably monitor animals moving through a heterogeneous landscape – a backdrop that includes areas of open grassland, woodland, and partial coverage.
“This type of work has been done before with whales, but of course the ocean is all blue, so counting is a lot less challenging,” Dr Isupova explained. “As you can imagine, a heterogeneous landscape makes it much hard to identify animals.”
African elephants were chosen for the study because they are the largest land animal and therefore the easiest to spot. However, Dr Isupova is hopeful that it will soon be possible to detect far smaller species from space.
“Satellite imagery resolution increases every couple of years, and with every increase, we will be able to see smaller things in greater detail,” she said. “Other researchers have managed to detect black albatross nests against the snow. No doubt the contrast of black and white made it easier, but that doesn’t change the fact that an albatross nest is one-eleventh the size of an elephant.”
Sign up to the E&T News e-mail to get great stories like this delivered to your inbox every day.