Reproducability code for Katsis, L. K. D., et al. (2022). "Automated detection of gunshots in tropical forests using convolutional neural networks." Ecological Indicators 141 doi: https://doi.org/10.1016/j.ecolind.2022.109128
This associated data used for model training and validation (749 gunshots, and 35239 background sounds collected in tropical forest sites in Belize) is archived on Mendeley Data doi: https://doi.org/10.17632/x48cwz364j.3
For detailed information and instructions on how to run predictions using this model, see the Opensoundscape docs.
Requirements:
- Python 3.7
- Opensoundscape
Classify files using the Google Colab script here
- Opensoundscape 0.7.1.
- User-friendly scripts to run the classification algorithm on your data are found in the subfolder Gunshot classification GUI
- This page contains instructions on running the scrips, and example data.
- Opensoundscape 0.7.1
- Scripts used to reproduce the manuscript are found in the subfolder Model and code
- This folder contains the training script, the final model, and the prediction script
- IMPORTANT - this model currently must be opened within Opensoundscape 0.5.0 or 0.6.0 to replicate results in paper. Results differ if opened in later versions of opensoundscape