Conserving Wildlife with Data Science
Some say we’re in our sixth mass extinction event, and in the face of climate change and declines in biodiversity, it’s important that conservation scientists have all of the tools they can get. Thankfully, data science has provided some increases in efficiency, as it has in many other fields.
For example, a program called “Wildbook” uses machine learning to identify wildlife quickly; in place of having a person identify each animal in a group individually, the software can go through a dataset of thousands of images of animals in a much shorter time. After that, if you need to find a specific animal in the group, the software can pick it out far faster than someone could by eye. When given an image, Wildbook first scans it for indications of wildlife, and then identifies unique markings on the animal to identify it with around 99% accuracy.
In combination with citizen participation and broad community efforts, data science methods like Wildbook can become even more powerful. For instance, in 2016, 350 experts, officials, and community members came together to collect over 40,000 pictures of zebras in Kenya. After that, it only took a few weeks for researchers to identify every zebra, thanks to the help of Wildbook and citizen scientists. Through the use of smartphone cameras and convenient apps, citizen-collected data is also being used to monitor illegal logging in the Amazon, for example. Citizens are even providing useful data unwittingly: one software system finds videos containing the phrase “whale shark” and scans them to identify any whale sharks that might be in them.
There’s no doubt that data science methods are more needed than ever in conservation biology; the impacts of global warming, poaching, and habitat loss are myriad, but hopefully new developments in the field can mitigate them.