Saved in:
| Main Author: | |
|---|---|
| Format: | Recurso digital |
| Language: | |
| Published: |
Zenodo
2024
|
| Online Access: | https://doi.org/10.5281/zenodo.10948975 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
Table of Contents:
- <p>This is the supporting data for: Clink, Dena J., et al. "Automated detection of gibbon calls from passive acoustic monitoring data using convolutional neural networks in the "torch for R" ecosystem." arXiv preprint arXiv:2407.09976 (2024).</p> <p>Link to GitHub Repository: https://github.com/DenaJGibbon/torch-for-R-gibbons</p> <p><strong>Summary of data and scripts.</strong></p> <p>To run the scripts download the data from Zenodo and add to your project directory.</p> <table style="border-collapse: collapse; width: 946pt;"><colgroup><col style="width: 236pt;"> <col style="width: 441pt;"> <col style="width: 269pt;"> </colgroup> <tbody> <tr style="height: 17.0pt;"> <td style="height: 17.0pt; width: 236pt;">R Script</td> <td style="width: 441pt;">Summary of Referenced Folders</td> <td style="width: 269pt;">Top-Level Data or Results Folders Referenced</td> </tr> <tr style="height: 35.0pt;"> <td style="height: 35.0pt; width: 236pt;">Part 1a. Variability benchmarking results</td> <td style="width: 441pt;">Training/test image folders (Danum, Jahoo, Combined); model run results for variability benchmarking.</td> <td style="width: 269pt;">data/training_images_sorted/<span>, </span><span>results/part1/</span></td> </tr> <tr style="height: 35.0pt;"> <td style="height: 35.0pt; width: 236pt;">Part 1b. Evaluate variability benchmarking results</td> <td style="width: 441pt;">Evaluation output folders.</td> <td style="width: 269pt;">results/part1/</td> </tr> <tr style="height: 35.0pt;"> <td style="height: 35.0pt; width: 236pt;">Part 2a. Train CNNs over multiple epochs</td> <td style="width: 441pt;">Training/test data for all sites; performance outputs for initial CNN evaluations (Cambodia, Malaysia).</td> <td style="width: 269pt;">data/training_images_sorted/<span>, </span><span>results/part2/</span></td> </tr> <tr style="height: 35.0pt;"> <td style="height: 35.0pt; width: 236pt;">Part 2b. Train CNNs (cont.)</td> <td style="width: 441pt;">Model output folders (binary & multi-class); rerun AUC updates; model type labels.</td> <td style="width: 269pt;">results/part2/</td> </tr> <tr style="height: 35.0pt;"> <td style="height: 35.0pt; width: 236pt;">Part 3a. Data augmentation</td> <td style="width: 441pt;">Augmented training images (Danum, Jahoo, Combined); test sets; model run outputs based on augmented data.</td> <td style="width: 269pt;">data/DataAugmentation/<span>, </span><span>data/training_images_sorted/</span><span>, </span><span>results/part3/</span></td> </tr> <tr style="height: 35.0pt;"> <td style="height: 35.0pt; width: 236pt;">Part 3b. Evaluation data augmentation</td> <td style="width: 441pt;">Outputs from evaluating models trained on augmented data; model labels by type.</td> <td style="width: 269pt;">results/part3/</td> </tr> <tr style="height: 35.0pt;"> <td style="height: 35.0pt; width: 236pt;">Part 3c. Eval. augmentation on different test set</td> <td style="width: 441pt;">Evaluation outputs for external test data (Maliau + Vietnam); multi-class model run outputs.</td> <td style="width: 269pt;">TestData/<span>, </span><span>results/part3/</span></td> </tr> <tr style="height: 35.0pt;"> <td style="height: 35.0pt; width: 236pt;">Part 4b. Comparison with BirdNET</td> <td style="width: 441pt;">BirdNET outputs for Jahoo, Danum, and Combined data; test output from jittered data augmentation.</td> <td style="width: 269pt;">results/part4/<span>, </span><span>results/part3/</span></td> </tr> <tr style="height: 35.0pt;"> <td style="height: 35.0pt; width: 236pt;">Part 5. Final model performance</td> <td style="width: 441pt;">Final evaluation images (per site); top-performing model folders; final performance CSVs.</td> <td style="width: 269pt;">data/CombinedImagesWAEvaluation/<span>, </span><span>results/part5/</span></td> </tr> <tr style="height: 18.0pt;"> <td style="height: 18.0pt; width: 236pt;">Part 6. Deploy Model over PAM data</td> <td style="width: 441pt;">Longer sound files (Jahoo); final trained model checkpoint (.pt file) used for inference.</td> <td style="width: 269pt;">data/WideArrayEvaluation/<span>, </span><span>results/part3/</span></td> </tr> <tr style="height: 35.0pt;"> <td style="height: 35.0pt; width: 236pt;">Part 7. Call Density Plots</td> <td style="width: 441pt;">Files for call density visualizations: GPS, selections, manually verified wavs or images (TP/FP).</td> <td style="width: 269pt;">data/calldensityplots/</td> </tr> </tbody> </table>