Find the accompanying manuscript published in Ecological Informatics here. To retrain the model, please see our Zenodo repository. Here, we stored all training and testing data, and explain how the user can retrain the model.

Installation

1. Clone git repository
2. In cmd, navigate to the repository. Change the path in the example command below.

cd d:\coding_projects\BatBuddy

3. Install environment using conda/miniconda

conda env create -n batbuddy --file environment.yml

4. Activate the conda environment

conda activate batbuddy

Using the tool

Analysing using UI

Simply open the tool in the command line:

python app.py

Parameter configuration is scarce in the UI, on purpose. If you're looking to change settings (like the overlap--as discussed in the paper, storing spectrograms, logging the analysis process in a csv file, or change the batch size), check out the next option:

Analysing using python interface

1. Edit the parameters at the end of the script (underneath if __name__ == "__main__":)

   • dir_list: Single path or list of paths.
   • log_path: False or a path where to store/find log file if you want to log the analysis (so the tool can continue later on where it left of).
   • files_per_batch: Number of recordings checked before writing to output file. The risk of setting this too high is an out of memory crash. If you only have a couple of GBs of RAM, set this at 1000. If you have more to spare, the default value of 5000 should be fine.
   • overlap: 0 when not using sliding window approach. 0.1-0.9 when using sliding window, where 0.1 if the proportion overlap between subsequent spectrograms analysed.
   • recursive: True if all dirs inside the specified dir(s) should be analysed. False if only recordings in the specified dir in dir_listshould be analysed.
   • proc: Number of logical processors to use to analyse recordings in parallel. This has been tested up until 12 processors, where runtime started leveling off around 8 processors. Results may vary on different machines.

2. Run the program in the command line:

python main.py