A.I. Recognizes Different Spider Webs
Pattern-recognition software was used to analyze images supplied by a spider expert.
Identifying animal species is the first, crucial step to any form of conversation. But it's a big job, especially if you're trying to identify spiders. Spiders occupy nearly every nook and cranny around the world - except for the polar regions - so they are good indicators for an ecosystem's health. But how to identify them?
Carlos Travieso at the University of Las Palmas de Gran Canaria in Spain and his colleagues have devised a computer systems that uses artificial intelligence to recognize webs. Travieso thinks that web construction is species specific. "Cobwebs are made by spiders whose genetics and morphology are all different from those of other species – and so their cobwebs are different, too. It's as distinctive as handwriting biometrics in humans," he told New Scientist.
To test that theory, Travieso used pattern-recognition software to analyze images supplied by spider expert William Eberhard of the Smithsonian Tropical Research Institute in Costa Rica. The software looked for unique features at a web's center and then correlated that information with details in the rest of the web. The software achieved a remarkable 99.6 percent accuracy in identifying the species of spider responsible for a web.
The team plans to use their A.I. system to count other species of animals and insects.
Credit: joruba / iStockPhoto