Animal experts can now identify, count, and describe species in the wild using cutting-edge artificial intelligence.
The technology, which is co-developed through data crowdsourcing, conserves wildlife while conducting its census.
The team led by Jeff Clune, a Harris associate professor at the University of Wyoming, gathered photographs of wild animals that are automatically collected by motion-sensor cameras.
They applied a technique called deep-learning system, which can identify up to 99.3 percent of the images. The accuracy of these images is verified by a team of 50,000 human volunteers.
"This technology lets us accurately, unobtrusively and inexpensively collect wildlife data, which could help catalyze the transformation of many fields of ecology, wildlife biology, zoology, conservation biology and animal behavior into 'big data' sciences. This will dramatically improve our ability to both study and conserve wildlife and precious ecosystems," Clune said.
Details of the study are published in the June 5 edition of the National Academy of Sciences.
Using Artificial Intelligence In Wildlife Census
Clune's team developed deep neural networks of images that mimic an animal's sensorial abilities. The process, however, is not an easy feat as it requires an extensive amount of data.
The images containing information about the species have to be tagged accurately for it to work properly. The data is sourced from Snapshot Serengeti, a citizen science project that distributed motion-sensor cameras to different parts of Tanzania. Millions of images of lions, cheetahs, elephants, and other animals in the wild are housed in this database.
"To better understand the complexities of natural ecosystems and better manage and protect them, it would be helpful to have detailed, large-scale knowledge about the number, location, and behaviors of animals in natural ecosystems," the authors wrote in the study.
For this data to become usable, Clune's team has to convert 3.2 million images into texts and numbers. Craig Packer, from the University of Minnesota, who is the head of Snapshot Serengeti, said that using Clune's technology allowed them to handle greater amounts of data.
Other Ways To Conduct Wildlife Census
Drones have become increasingly useful in measuring wildlife data. In a study published February 13 in Methods in Ecology and Evolution, researchers reported the results of an experiment called the #EpicDuckChallenge.
The task is to deploy thousands of plastic duck replicas on an Adelaide beach in South Australia. The researchers found that drones are capable to produce more accurate data on wildlife population compared to traditional census methods.