Image recognition software for plants

Quick and accurate identification of a particular type of plant can be difficult, especially if you’re not a botanist and you don’t carry around pictures of all 250,000+ species on Earth. Thanks to a computer program developed at Edith Cowan University (ECU), plant classification could become as simple as taking a photo.

Computer science researcher Oluleye Hezekiah Babatunde is working on a program that uses high-tech image recognition software to identify a plant based on a digital photo of a single leaf. His work has been described in the Journal of Agricultural Informatics.

“The program uses a combination of highly complex mathematical algorithms and artificial intelligence to identify leaves based on their colour, shape and texture,” Babatunde said.

“In testing so far, the program correctly identifies leaves more than 93% of the time.”

The software could have extensive applications in the agricultural sector, according to Babatunde, who said the same processes used to distinguish between different species could be used to monitor damage to crops by pests, weather and herbicide.

“There’s potential for farmers to use this tool in an app form to get real-time updates on their crops with a simple smartphone image,” he noted.

Babatunde is now seeking partnerships to take the project out of the lab and into the field.