Skip to content

Model Card: Bird Species Model (ppq7zxqq)

This document provides details about the pre-trained EDANSA model specifically tuned for North American arctic bird species.

Model Details

  • Model ID / Run Name: ppq7zxqq
  • Description: A bioacoustics model trained to detect specific bird species commonly found in North American arctic environments.

Assets Location

The model weights, configuration file, recommended thresholds, and raw predictions for threshold tuning are located within this repository:

  • Model Weights: assets/ppq7zxqq/best_model_200_val_f1_min=0.7857_ppq7zxqq.pt
  • Configuration File: assets/ppq7zxqq/model_config_ppq7zxqq.json
  • Recommended Thresholds (Test Set): assets/ppq7zxqq/ppq7zxqq_thresholds.csv (Contains optimal F1 thresholds derived from the test set - recommended starting points)
  • Raw Predictions (for custom thresholding): assets/ppq7zxqq/all_predictions_epoch_200_ppq7zxqq.csv (Contains raw model outputs/logits. These need transformation, e.g., via a sigmoid function, to be interpreted as probabilities between 0 and 1. Useful for advanced users wanting to determine custom thresholds).

Performance Metrics (Test Set)

The Short Label corresponds to the prefix used in output CSV column headers (e.g., pred_LALO).

Label (Species) Short Label Test AUC Test F1 Score F1 Threshold
Lapland Longspur LALO 0.974 0.909 0.433
White-crowned Sparrow WCSP 0.981 0.864 0.635
American Tree Sparrow ATSP 0.991 0.835 0.222
Willow Ptarmigan WIPT 0.942 0.802 0.701
Savannah Sparrow SAVS 0.938 0.797 0.578
Common Redpoll CORE 0.989 0.786 0.197

Metric Definitions:

  • AUC (Area Under the ROC Curve): Measures the model's ability to distinguish between classes across all possible thresholds. A value closer to 1.0 indicates better separability.
  • F1 Score: The harmonic mean of precision and recall at the specified F1 Threshold. It provides a balance between finding relevant sounds (recall) and ensuring the found sounds are indeed relevant (precision). A value closer to 1.0 is better.
  • F1 Threshold: The confidence score threshold optimized on the test set to achieve the reported F1 Score. Use this as a starting point for thresholding predictions.

Intended Use

This model is intended for detecting specific bird species in passive acoustic monitoring data from relevant geographic regions (e.g., North American arctic/sub-arctic). It can be used for species occurrence studies or as input for further analysis.

Limitations & Thresholding

  • Performance may vary depending on the specific acoustic environment, recorder type, noise levels, distance to the sound source, and regional variations in vocalizations.
  • The model's performance on species not included in the training data is unknown.
  • Threshold Adjustment: The optimal confidence score thresholds provided in the ppq7zxqq_thresholds.csv file are recommended starting points. Users should evaluate and potentially adjust these thresholds based on their specific project goals (e.g., prioritizing recall over precision) and test data.
  • Advanced Thresholding: For users wanting to perform detailed threshold analysis or optimize using different metrics, the raw model predictions (logits, which need transformation like sigmoid to become 0-1 probabilities) are available in all_predictions_epoch_200_ppq7zxqq.csv.