Jan 17, 2013
American learned how to recognize animals
Harvard University Biologist Walter Scheirer developed an algorithm, that defines an animal kind when you take photos of animal on a digital camera, reports New Scientist. Technology is planned to be used in smartphones applications, and also as assistance to scientists working with automatic cameras-traps.
How does it work, this «zoorecognizer»?
The camera makes a series of photos and its processing in device progress emits the «dynamic» group of pixels, which was appeared or moved in the scene.
Further dynamic group of pixels checks -according to a set of attributes - with «familiar» images of animals, included in the database.
So, being under testing the application distinguished three types of squirrels in 85% of cases.
The algorithm can be used in the work of automatic trap-cameras reacting to movements. Now experts have to sort manually tens or hundreds of photos, made in blind way; however, due to a new algorithm, equipment can be configured so as to preserve only the photos you need.
Trap-cameras are usually used when photographing an animal is dangerous or difficult. For example, it can be used in the case of a major predator of the cat family (leopard, tiger, ounce, lynx and jaguar) and small nocturnal animals shooting.
Meanwhile…
Fujifilm prepared for launch a new model of a compact digital camera FinePix X700EXR, which implements some of new features.
In addition to the already traditional face-recognition mode, modes of animals recognition are available. So, it supports the ability to recognize up to 10 cats or dogs in one frame.