Even wildlife experts have difficulty telling a lynx and a bobcat apart from many of the images captured in the less populated parts of the Canada. Yet telling the animals apart is important for conservation.
While the trained eye may sometimes struggle with image determination, artificial intelligence appears more adept at interpreting images if wildcats. This development comes from the University of British Columbia.
The effectiveness of the artificial intelligence has been shown in a new study. The study has assessed various wildcat images, many of them submitted through a citizen science project. Such projects show that while images of wildcats can be produced, the differentiation process is often challenging.
According to one of the researchers, TJ Gooliaff, in conversation with Phys.org: “Camera-trapping and citizen-science studies collect many wildlife images for which correct species classification is crucial.”
Yet accuracy of interpretation is also important, as Gooliaff explains: “Even low misclassification rates can result in erroneous estimation of the geographic range or habitat use of a species—including underestimation of the occupancy, habitat preferences or distribution of a species. This potentially hinders conservation and management efforts.”
The technology not only applies to wildcats for there are other animals that appear similar when studied at a distance, such as with bears, deer, lemurs, and antelopes. This is due to similarities of size, shape or colour. To add to this there are issues when images are not all that clear, such as with reduced lighting or where the image if fuzzy.
To assess the artificial intelligence the researchers used 4,399 images of bobcats and lynx taken by citizen scientists throughout British Columbia. The aim was to evaluate the provincial distribution of each species. The success of the artificial intelligence was compared against the opinions of 27 different experts.
This process was used to further train the artificial intelligence, with the system focusing on aspects of the cats like paws, head, and tail.
The description of the artificial intelligence and the way it was trained is outlined in the Journal of Wildlife Management. The research paper is titled “Measuring agreement among experts in classifying camera images of similar species.”
