When states want to assess quail populations, the process can be exhausting, time-consuming and expensive.
That means spending hours in the field listening to calls. Or leave a recording device in the field to pick up the sounds made, only to spend hours later listening to that audio. Then repeat this process until there is enough information to start making population estimates.
But a new model developed by researchers at the University of Georgia aims to streamline this process. Using artificial intelligence to analyze terabytes of quail call records, the process gives wildlife managers the ability to gather the data they need in minutes.
“The model is very accurate, picking up between 80% and 100% of all calls, even in the loudest recordings. So you can take a recording, put it in our model and it will tell you how many quail calls that the recorder heard,” said James Martin, an associate professor at the UGA Warnell School of Forestry and Natural Resources who worked on the project, in conjunction with the Georgia Department of Natural Resources, for about five years. “This new model allows you to analyze terabytes of data in seconds, and what this will allow us to do is increase surveillance, so you can literally put hundreds of these devices and cover a lot more surface and do it with much less effort than in the past.
The software represents approximately five years of work by Martin, postdoctoral researcher Victoria Nolan, and many key contributors who worked with a code writer to create the model. It’s also part of a bigger shift in wildlife research, where computer algorithms are now helping to do work that once took humans thousands of hours.
Increasingly, computers are getting smarter, for example, to identify specific noises or certain features in photos and sound recordings. For researchers like Martin, this means that hours spent on tasks such as listening to audio or viewing footage from game cameras can now be done by a computer, freeing up valuable time to focus on other things. other aspects of a project.
The new tool can also be a valuable resource for state and federal agencies looking for information about their quail populations, but with limited funds to spend on a given project. “So I think it’s something that states could skip over replacing their current surveillance with acoustic recording devices,” Martin added.
The success of the software was recently documented by the Journal of Remote Sensing in Ecology and Conservation.
As the software is used more and more and exposed to sounds from new geographies, Martin said, it becomes even smarter. As it stands, quails offer several different types of calls. But when the software is exposed to a variety of sounds that aren’t quail, he says, it’s better able to distinguish correct calls from the ambient sounds of the grass and trees around them.
Over time, the software will become more demanding.
“That’s why you have to keep giving it training data, and as you move geographically, you come across new sounds that you haven’t trained the model for,” he added. “It’s always a question of adaptation.”
Remote sensing in ecology and conservation
The title of the article
The development of a convolutional neural network for the automatic detection of Northern Bobwhite Colinus virginianus covey calls
Publication date of articles
August 24, 2022
Disclaimer: AAAS and EurekAlert! are not responsible for the accuracy of press releases posted on EurekAlert! by contributing institutions or for the use of any information through the EurekAlert system.