Pennsylvania State University paleobotanist Peter Wilf and colleagues have developed new software for identifying families of leaves.
Botanists currently rely on a manual method of identification that has not changed much since it was developed in the 1800s. The lead architecture method is a painstaking process that involves using an unambiguous and standard set of terms from a big reference book to describe leaf form and venation, and correctly identifying a single leaf's taxonomy can take two hours.
The new software, which Wilf developed with Brown University neuroscientist Thomas Serre, combines computer vision and machine-learning algorithms to identify families of leaves in only milliseconds.
Wilf views the tool as an assistant, considering its accuracy rate for identifying patterns in leaves and linking them to the families they potentially evolved from is 72 percent.
Wilf and Serre have fed the program 7,597 images of leaves that have been chemically bleached and then stained. Once the software processes these ghost images, it creates a heat map on top of them, with red dots pointing out the importance of different codebook elements, or tiny images illustrating some of the 50 different leaf characteristics. Together, the red dots highlight areas relevant to the family to which the leaf may belong.
Wilf wants to feed the software tens of thousands of images of unidentified, fossilized plants.
View Full Article
Abstracts Copyright © 2016 Information Inc., Bethesda, Maryland, USA
No entries found