OREGON STATE UNIVERSITY

Insect identification website takes wing at OSU

08/02/2011

CORVALLIS, Ore. – A new online tool being developed at Oregon State University uses interactive pattern recognition technology to help researchers quickly and accurately identify species of moths and butterflies.

Jeff Miller, a professor in OSU's College of Agricultural Sciences, and colleague Hans Luh, a senior research associate with the university's Integrated Plant Protection Center, are the creators of the tool, called Lepidoptera Wing Pattern Identification System, or LepWing ID. This pilot project allows users to compare a digital image of a specimen against a library of more than 1,600 photos.

Those images are just a fraction of the roughly 150,000 species of moths and butterflies that have been identified worldwide, but LepWing ID (http://ipmnet.org/lepid) still represents a significant improvement in identification. Compared to thumbing though a paper field guide to put a name to a particular species, the system offers a much faster and more accurate way of identifying moths and butterflies, Miller said. It is the first free, digital tool for identification available to both scientists and the general public.

Though still in its early phases, LepWing ID will be a valuable resource for scientists, as the winged insects are an indicator species of overall ecosystem health. Once the library is expanded, the system also has the potential to be used for non-research purposes. For instance, an agricultural inspector who finds a moth in a load of imported fruit could look up the insect in LepWing ID and determine if it is new to the region, and whether it is a pest or a beneficial.

"It would expedite the response to finding a new insect, perhaps by as much as one whole season, which could make a big difference in responding with appropriate management tactics," Miller said.  

LepWing ID works by matching a color pattern on a specific section of the wing to the same wing section in library images. Users also have the option of selecting certain traits, such as the dominant color of the specimen, from a menu to narrow down possible matches. Results are displayed in descending order, with the most probable matches displaying first. Because there are multiple photos of most species, the system might return the same species several times. But that's a good thing, Miller said, because it reassures the user that the results are accurate.

Users don't even need to have the specimen in hand to use the system: they can upload a computer-made illustration or even a hand-drawn cartoon of what they've glimpsed and LepWing ID will be able to search for potential matches, Miller said.

Miller took many of the photographs of moths and butterflies that make up the system's library, and he continues to add images to increase the tool's accuracy. He hopes LepWing ID will someday have tens of thousands of images to match against an uploaded photo of an unknown species.

Because LepWing ID is free, it's also a tool for the casual gardener or naturalist who is curious about the butterfly out in the garden.

In the future, Miller and Luh envision the LepWing ID model could be used to identify species of stink bugs, beetles, ticks, bees – even plants.

"I think the use is wide-reaching in biology," Miller said.