Birds With Labels Model¶
The Birds-With-Labels model combines all five zenodo datasets mentioned in the introduction. It therefore utilized almost 407 hours of soundscape recordings with 155,584 bounding box labels spanning across 282 different bird species.
Formerly All-In-One
Until 28 October 2026 this model was called All-In-One. Files made before then still carry the old name: rename All-In-One.pt to Birds-With-Labels.pt and pass --species-mapping Birds-With-Labels. BirdBox no longer accepts the old name.
More on Birds-With-Labels:
- Training: training curves and boxes per species and split.
- Model Types: the export formats and how closely they match the PyTorch model.
Precision, Recall and F1-Score¶
The graph below shows the performance of the model on the test split under different confidence thresholds: precision, recall and the F1-score, each averaged over the species (macro average). To optimize the F1-score, pick a confidence threshold around 0.27.
Rising Recall at Low Confidence
The rising recall from confidence values 0 to ~0.1 is unusual but by design. At very low confidence thresholds, the merging algorithm produces imprecise merged boxes that can accidentally overlap more ground truth labels. See How it works for details.
Top Performing Species¶
The figure shows the F1-score against the confidence threshold for the 12 best species among the 82 with at least 50 labels in the test split. Species with fewer labels are left out: with only a few labels, a high F1 is mostly chance. Three species reach an F1 of 1.0 on two to four test labels. At a confidence threshold of 0.4, all 12 species reach an F1 of at least 0.79.
