|A mass migration of wildebeests in Serengeti National Park, Tanzania, Africa. / Photo by: Daniel Rosengren via Wikimedia Commons|
A team of researchers employed artificial intelligence to identify, count, and describe wild animals. The software was able to identify the animals with more than 90 percent accuracy.
The AI supports deep neural networks, a form of computational intelligence, to detect wild animals in image samples. To train the software, the research team obtained relevant data from Snapshot Serengeti, a citizen science project on the Zooiverse.org platform. The science project has millions of photos of wild animals such as cheetahs, elephants, and lions that were captured in their natural habitat via motion-sensor cameras.
“We wanted to test whether we could use machine learning to automate the work of human volunteers. Our citizen scientists have done phenomenal work, but we needed to speed up the process to handle even greater amounts of data,” stated Craig Packer, a co-author of the study and head of Snapshot Serengeti.
Each image from the vast collection of the project was tagged and labeled accurately to train the software. They also included notable details like the type of species of each animal, the number of animals present in a picture, and other items needed by the AI. Their efforts were aimed to provide human scientists with an automated system that would allocate information from images.
After training the AI system, the researchers tested its accuracy in describing the animals in each photo. With DNN, the AI was able to automate identification by up to 99.3 percent.
“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,” explained Jeff Clune, the senior author of the study and an associate professor at the University of Wyoming.
Their technology would be useful for animal experts who regularly study matters related to the conservation and ecosystems of wild animals. Also, the system could be applied for cataloging archived data.