Bear research made easier

February 09, 2006, vol. 35, no. 1
By Barry Shell



Document Tools

Print This Article

E-mail This Page

Font Size
S      M      L      XL

Related Stories

Wildlife biologists have it tough. They frequently must endure harsh conditions observing their subjects.

Then comes the tedium of transporting and analysing reams of data. A group of SFU computer scientists wants to change that with new computerized video technology.

Shelley Marshall, a graduate student studying grizzly bears, is one of their first customers. Last year the Yukon government and local First Nations approved a new form of tourism called wilderness bear viewing. A guide brings four tourists to a riverbank only metres away from wild grizzly bears feeding on salmon.

For her master's thesis, Marshall is studying whether such tourism will affect the bears.
Marshall, who began the study last year, was not looking forward to the boring task of watching for bears in sub-zero temperatures. She also had to cover a large area, and she didn't want the bears to know they were being watched in case they altered their natural activities.

“We were scratching our heads, trying to figure out how to do this, when an e-mail came from Greg Mori,” says her supervisor Kristina Rothley, an assistant professor in the school of resource and environmental management.

Mori, an assistant professor in computing science, asked whether anybody wanted to collect digital video on animals behaving in a natural forest setting. Mori's specialty is computer vision. His software can recognize objects and activities on video. He needed subjects for a new research initiative called scientific data acquisition, transportation and storage (SDATS).

Mori had partnered with Richard Vaughan and other computing science professors specializing in robotics, data mining, and data transmission. “We want to help natural scientists by automating the most tedious aspects of their job,” says Mori. The SDATS system can unobtrusively watch wild animal habitat 24 hours a day, seven days a week. Marshall obtained 200 hours of video. Segments containing bears totaled only one hour.
This summer Mori will run his software on the video data to see if it can see bears as well as humans.

Search SFU News Online