Massachusetts Institute of Technology researchers led by Sai Ravela have developed SLOOP, software designed to partly automate the matching of images of animals in massive catalogs.
The SLOOP system mines thousands of images using pattern-recognition algorithms to analyze features such as a specimen's stripe or spot configuration, and then identifies an average of 20 most probable matches for an individual. The system also uses feedback from online users asked to choose the most similar pair of animals so that it can reorder and pare down the list.
The researchers say the combination of computer-vision algorithms and crowdsourcing enables rapid identification of image matches among thousands of photos with 97-percent accuracy.
Ravela notes that most pattern-recognition algorithms lack the sophistication to parse out the complexity in animal patterning. His team has devised multiple algorithms to identify matching patterns, including some that adjust for changes in an animal's lighting, orientation, and geometry, and other algorithms that overlay images, comparing stripe or spot positioning.
New Zealand's Department of Conservation currently uses SLOOP to track threatened skink populations, and the system may ultimately help researchers gain broader insights about animal behavior, such as species' migration patterns and breeding habits.
From MIT News
View Full Article
Abstracts Copyright © 2013 Information Inc., Bethesda, Maryland, USA
No entries found