Running the workflow

To run the entire workflow, use:

pamflow run

This is equivalent to kedro run — pamflow is built on top of Kedro, so both commands work interchangeably. This documentation uses pamflow run.

The complete workflow runs, in this order of dependencies, filter_media, data_preparation, graphical_soundscape, acoustic_indices, species_detection and quality_control. Two pipelines are not part of it and must be run explicitly: detection_validation, which needs the manual annotations reviewed by an expert, and export.

For better control and to understand each processing step, it is often useful to run individual pipelines:

pamflow run --pipeline <pipeline_name>

The valid names for --pipeline are:

Pipeline

What it does

In pamflow run

data_preparation

Standardizes recordings and field metadata into the pamDP format.

✅

quality_control

Timeline, locations, survey effort, audio outliers and timelapses.

✅

graphical_soundscape

Graphical soundscapes.

✅

acoustic_indices

Acoustic indices per deployment.

✅

species_detection

Species detection with BirdNET, audio segments and annotation files.

✅

detection_validation

Per-species thresholds from reviewed annotations.

export

Converts the pamDP outputs to the GBIF (camtrapDP) format.

pamflow

The complete workflow (same as the default).

✅

graphical_soundscape, acoustic_indices and species_detection include the filter_media step, so they can be run on their own once data_preparation has been run. See the pipeline details for the inputs, outputs and parameters of each one.

To work on a subset of the deployments, list them in deployments_filter, see Filter media.

A pipeline consists of multiple nodes, where each node represents a single processing task (for example, validating metadata, generating spectrograms, or running species detection).

To execute a specific node:

pamflow run --nodes <node_name>

To list the available pipelines and see the nodes of a pipeline, use the Kedro registry (no additional installation needed):

kedro registry list
kedro registry describe <pipeline_name>

For example, kedro registry describe species_detection prints the node names of that pipeline. The pipeline documentation lists the inputs, outputs and parameters of each node.

For a complete list of execution options, see the Kedro command reference