Passive acoustic monitoring observations of amphibians in Ukrainian Roztochya

This dataset contains amphibian observations collected through passive acoustic monitoring in the Ukrainian Roztochya region. Data were recorded using autonomous acoustic recorders and analyzed using a custom deep learning model for amphibian species identification, developed based on the BirdNET architecture ( doi.org/10.5281/zenodo.17736511 ). The dataset represents observations aggregated to unique combinations of hour/date and location (one record per species per location per date per hour).

Original acoustic data consisted of continuous recordings that were processed through the custom-trained model for automated species detection. To ensure data quality, automated detections were randomly sampled and verified by human experts. An optimal confidence threshold for species identification was determined using logistic regression following the methodology of Wood and Kahl (2024; doi.org/10.1007/s10336-024-02144-5 ).