Discipline: Computer Sciences and Information Management
Subcategory: Computer Science & Information Systems
Olatide Omojaro - Georgia Institute of Technology
Co-Author(s): Kathryn Hollowood, Wesleyan College, Rochester, NY Dalwindeerjeet Kular and Eraldo Ribeiro, Florida Institute of Technology, Melbourne, FL
Biodiversity conservation is a major concern of modern society. Pressures such as littering, pollution, and climate change deteriorate the natural habitats and reduce biodiversity. To help repair this issue, we need some way to track these habitat damages. We need a bio-indicator. Frogs make excellent bio-indicators due to their permeable skin. This skin allows them to absorb water and Oxygen. However, when they live in polluted areas they also absorb toxins. As pollution rises, some frog populations suffer decline. Citizen scientists have been monitoring these frog populations by recording their calls for years. However, this process requires expensive equipment, training, and a sizable time commitment. This paper focuses on the automatic classification of frog calls using computer programs. Features were extracted from the audio data and classified using two classification methods: Locality Sensitive Hashing and Gaussian Mixture Modeling. Tests performed on a dataset of frog calls of 15 different species produced promising classification results for the Gaussian mixture model approach.
Funder Acknowledgement(s): National Science Foundation (NSF) Grant No. 1263011 and 1152306
Faculty Advisor: Eraldo Ribeiro,