BirdNet Custom Training Pipeline Setup Instructions
Prerequisites
- Access to the shared Azure Machine Learning (AML) workspace.
- Datastore: The following datastore must be available:
customaudiodata
: Stores the training data and the output custom BirdNet model.
Steps for Setting Up the Pipeline in Azure Machine Learning Studio
1. Log into Azure Machine Learning Studio
Navigate to Azure Machine Learning Studio and ensure you’re using the correct workspace.
2. Verify Datastores
Ensure that the necessary datastore, customaudiodata
, is correctly set up.
3. Verify Pipeline Configuration
- The pipeline is named
BirdNet-Custom-Training-Pipeline
. - The pipeline uses a GPU cluster to train the BirdNet model.
4. Running the Pipeline
- Open the pipeline in AML Studio.
- Ensure that the training data is correctly stored in the
customaudiodata
datastore (BirdNET_training_datasets/Stage1/Training/
). - Set the output location in
customaudiodata
for storing the trained BirdNet model. - Submit the pipeline and monitor the run in Experiments.
5. Monitor Execution
The pipeline’s progress can be monitored in the Experiments tab. Logs and results will be available for each step.