Machine learning tools for discrimination of sounds produced by different insect species in individual trees

On-Demand

Presenting Author(s)

Richard W. Mankin

Research Entomologist
USDA-ARS
Gainesville, Florida

Co-Author(s)

SS

Sruthi Sentil

University of Florida
Tallahassee, Florida
SM

Seth McNeill

Associate Professor, Computer Engineering
Embry Riddle University
Prescott, Arizona

Insects such as red palm weevil that feed hidden inside trees can be detected by acoustic methods but, if multiple species are present in a tree, the sounds of each species are difficult to distinguish from each other, Previously the sounds have been discriminated by distinguishing differences in the sound patterns of chewing and movement of each species. We tested a hypothesis that statistical, machine learning methods can be applied to automate the process of identifying differences among insect pest species in the same tree. Testing was successful, and the results suggest that such methods also can be applied to other agricultural systems in which multiple species hidden in individual trees must be discriminated from each other.