Ring any bells? Readers of the Approach may remember a 2014 post on the Image-Based Ecological Information System, or IBEIS, an experimental software tool that can identify individual animals pictured in the thousands of ordinary photos taken by tourists in game parks such as Yellowstone National Park, Amboseli National Park in Kenya, or the Galapagos Islands off the coast of Ecuador. The 2014 post explains how the system works.
As the poster suggests, a test of the system (a collaboration between researchers at Rensselaer Polytechnic Institute, the University of Illinois at Chicago, Princeton University, and the wildlife conservation organization Wild Me) will be held March 1 and 2. Chuck Stewart, head of the Department of Computer Science at Rensselaer, and a lead researcher on the project, gave this quickie description of how the event will work:
Cars will sign up in advance and enter the park at pre-specified times. We will give them instructions, test their cameras, give them GPS dongles, and send them on routes we determine. They are supposed to take pictures along their route of every zebra and giraffe they see on the left side of their car. When they are done we will extract the pictures from their SD cards on their cameras, run the software and show results. There are a lot of logistics to doing this, especially planning and implementing the collection software in four weeks from 8,000 miles away.
Stewart said Rensselaer students developed the core algorithms that drive the system and have been working diligently to develop the collection software. In particular, he credits Jason Parham with writing the core IBEIS code that finds animals in each photo (he is also leading the team that is writing the data collection software), and Jon Crall with writing the core IBEIS individual animal identification code (in other words – given that the picture shows a zebra, which zebra is it?)
Parham and Stewart will be on scene in Nairobi for the test. You can follow the Great Zebra & Giraffe Count on their Facebook page.