Students at an Edmonton post-secondary institution are using artificial intelligence (AI) to help the provincial museum unscramble avian mysteries. Instructors with NorQuest College’s machine learning analyst program caught wind of the needs of researchers at the Royal Alberta Museum who were trying to more efficiently identify wild eggs donated to it. Corey Scobie, the museum’s assistant curator of ornithology, said the project using AI came through conversations with the instructors. He said it usually takes him and a volunteer researcher two to three hours to identify the eggs through conventional means. “There are all these small perching birds that lay eggs that are very similar, so some sort of tool that, where if we could take a picture and at least narrow down what group it belongs to, whether it’s pipits or warblers or sparrows, and maybe even the species, would have been super helpful,” Scobie told CTV News Edmonton on Friday. Scobie brought up his quandary with instructors from NorQuest who sometimes work at the museum, leading to a collaboration in which students in the machine learning analyst program helped him identify 21 wild eggs. The students spent hours taking photos of the eggs and building an app that uses AI to identify which bird species the eggs belong to. Student Anchonia Offor said her group traded several ideas – “a lot of back and forth” – figuring out how their model would be able to learn the data based on information they gathered from Scobie and his research process. “This model learns from the different pixels,” Offor said. “It’s able to learn exactly what we would learn manually, so the size, the colour (and) the different markings on the bird egg, it would learn all of that using the pixels.” The app is still a prototype. Once it’s complete, it could help the museum identify an egg in minutes. Scobie said there’s “definitely interest” among museums for time-saving tools such as the NorQuest app. “We’ll constantly need to be checking the results of these tools, but while we do that, we’re also improving the model,” he said. “There’s big potential in museum curators using technology like this to help identify (material), or at least save time.” With files from CTV News Edmonton’s Sean McClune
