Test Mode

Test Mode#

Overview#

In paidiverpy, test mode allows users to execute pipeline steps in a diagnostic mode using the InvestigationLayer class. This is especially useful during debugging, development, and understanding the behavior of specific processing steps.

When enabled, test mode runs a pipeline step without altering the input images. Instead, it generates diagnostic outputs such as plots or logs to visualize the operation of the step.

This mode is beneficial for:

  • Understanding how a specific step operates

  • Non-destructive testing of pipeline steps

  • Visualisation of algorithmic effects before application

  • Debugging support through detailed plots and metadata logging

Enabling Test Mode#

To activate test mode, set the test parameter to True within the relevant pipeline step in the configuration file. For example:

general:
  # some general inputs

steps:
  - colour:
      name: 'colour_correction'
      mode: 'grayscale'
      test: True

In this configuration, the colour_correction step will run in test mode, generating logs and diagnostic plots without modifying the input images.

How Test Mode Works#

When test: True is specified, the step is handled by the InvestigationLayer class instead of the standard processing class. This class:

  • Loads metadata and images but skips transformation

  • Produces plots such as:

    • Original vs. resampled image locations (plot_trimmed_photos)

    • Brightness histograms (plot_brightness_hist)

    • Polygon overlays (plot_polygons)

  • Writes all plots to a dedicated subfolder within the output directory

The type of plot that is generated depends on the step type.

Example Output Structure#

When run with test: True, the pipeline creates a directory named after the step, such as:

output-path/number_of_step-step_name/
    └── 01_colour_correction/
        ├── graph_trimmed_images.png
        ├── histogram_brightness.png
        └── polygons_overlay.png

Where output-path is the specified output_path in the configuration file, number_of_step is the order of the step in the pipeline and step_name is the name of the step.

Additional Considerations#

  • Output Path: If the output_path is remote, the InvestigationLayer will skip plotting and issue a warning.

  • Metadata Requirements: Some plots require specific metadata fields (e.g., image-longitude, image-latitude, and others). If these fields are missing, the corresponding plots will not be generated and a warning will be issued.

  • Gallery Examples: Explore example notebooks with test mode enabled in the Gallery.