Goals: To adapt object detection domain adaption techniques developed by the Beery lab at MIT for use in acoustic detection of species. The project focuses on developing methods to analyze bioacoustics data for biodiversity monitoring, with particular application to secretive marsh bird species in Pennsylvania.

Overview: The ANTENNA project has developed a pipeline for converting acoustic data to work with existing object detection models. Preliminary models have been trained and compared against existing benchmarks as part of this first project year. The work supports two related field studies: monitoring King Rail populations in northern Pennsylvania and analyzing dine-scale habitat use of secretive marsh birds. Future work will replace the existing model’s feature extraction backbone with a stronger baseline mode, followed by retraining and evaluation. This research contributes to the Global Center’s broader aims of developing AI-enabled approaches for biodiversity monitoring.