Submit for Inference

This script is used to submit a single image for inference against a trained model.

usage: submit_for_inference.py [-h] [--experiment_name EXPERIMENT_NAME]
                               [--model_id MODEL_ID] [--image_file IMAGE_FILE]
                               [--settings SETTINGS]
                               [--download_folder DOWNLOAD_FOLDER]
                               [--keep_upload_folder [KEEP_UPLOAD_FOLDER]]
                               [--cluster CLUSTER] [--use_dicom [USE_DICOM]]

Named Arguments

--experiment_name

Name of experiment the run should belong to

Default: “model_inference”

--model_id

Id of model, e.g. Prostate:123. Mandatory.

Default: “”

--image_file

Image file to segment, ending in .nii.gz if use_dicom=False, or zip of a DICOM series otherwise. Mandatory.

--settings

File containing Azure settings (typically your settings.yml). If not provided, use the default settings file.

--download_folder

Folder into which to download the segmentation result. If this is provided, the script waits for the AzureML run to complete.

--keep_upload_folder

Whether to keep the temporary upload folder after the inference run is submitted

Default: False

--cluster

The name of the GPU cluster in which to run the experiment. If not provided, use the cluster in the settings.yml file

Default: “”

--use_dicom

If image to be segmented is a DICOM series and output to be DICOM-RT. If this is set then image_file should be a zip file containing a set of DICOM files.

Default: False