This decommissioned ERA site remains active temporarily to support our final migration steps to https://ualberta.scholaris.ca, ERA's new home. All new collections and items, including Spring 2025 theses, are at that site. For assistance, please contact erahelp@ualberta.ca.
- 1256 views
- 2367 downloads
Testing the Accuracy of a birdNET, Automatic bird song Classifier
-
- Author(s) / Creator(s)
-
In recent years, automated bird song classification programs have been becoming more common among
researchers as a way to study, track, and monitor birds. In our research, we tested the accuracy of one such
program called BirdNET. We tested 225 recordings by uploading them to BirdNET and manually classifying them
to see how often BirdNET was accurate. The overall accuracy of BirdNET was 91.5%, with this number increasing
when it came to bird songs that BirdNET was more familiar with, and dropping when it came to other bird songs
that BirdNET was unfamiliar with. This paper will explore why such a program is needed, how it can be helpful to
biologists, researchers, and anyone else interested in or looking to learn more about bird songs. This study also
includes the methods used to test BirdNET, discussion about how automated bird song recognition programs
can be improved, limitations when it comes to automated bird song recognition software, and other relevant
studies about acoustic monitoring and automatic bird recognition programs. -
- Date created
- 2020-07-31
-
- Type of Item
- Conference/Workshop Poster