Kilimanjaro Lion Conservation Project

Category: General | Date: Jun 30 2008 | By: Seamus

Automated pattern recognition has been something of a grail for wildlife researchers wanting to sample large populations of animals. The first I heard of this was in connection with some cheetah work in Tanzania. It was rumoured to be fairly expensive, but I thought it intriguing [every cheetah has a unique spot pattern, like a fingerprint]. Sometime in the past few months I had coffee with a slightly wild but brilliant acquaintance of mine who studies zebras. We got to talking about how efficient it would be to simply record zebra stripe patterns and then apply some kind of wickedly-complex mark-recapture stats to come to come up with estimates of population density. The only tools required: a camera, some fancy software and a formidably mathematically-inclined brain. No collars or unique ID marks necessary. By the time I had finished my coffee (they make ‘em large and strong here in Kenya) I had established that he had a prototype going, but it requires careful manual entry of the images, and the user needs to show the software where to look for the parameters. In other words, its still fiddly.

This is why I was excited to learn that some folks working at the place of Mandela’s incarceration, Robben Island, have got it right. They describe pretty much an automated system for penguin image capture and mark-recognition. Click here to read all about it.

(photo from PJ Barham / BBC)

You might ask how this applies to lion conservation. Well, I haven’t thought of an answer to this yet. But its an important development, which I think could be almost as significant as the advent of the GPS collar in wildlife monitoring. Of course, lions are uniquely identifiable (I’ve written about this before on this blog) but automated pattern recognition might be more applicable to surveying what lions eat, than lions themselves.

Technorati : conservation, cool science, lion research, lions, pattern recognition, penguins

Share This