AI is helping wildlife biologists identify rare beasts on the Serengeti
New research has shown that an algorithm can be as effective as humans at identifying animals.
Some background: Motion-sensitive cameras automatically photograph wild animals, providing massive amounts of data, but humans have to analyze every image.
The news: AI was able to accurately catalog 3.2 million photographs from the Snapshot Serengeti project with 96.6 accuracy—matching the record set by human volunteers. The software could save an estimated 17,000 person-hours worth of remaining work.
Why it matters: This success could propel fields like ecology, zoology, and biology to become “big data” disciplines. “This is a game changer for wildlife ecology," says Craig Packer, head of the Snapshot Serengeti project.
Image credit:
- Proceedings of the National Academy of Sciences