Workflow

Module dedicated to workflow management. It implements the Workflow class that offers high-level features based on a C++ implementation. You will be able to create, modify, load and run workflows composed by built-in Ikomia algorithms or any of those available in Ikomia HUB.

Import

from ikomia.dataprocess import workflow

Classes

Workflow([name, registry])

Workflow management of Computer Vision tasks.

Functions

create([name])

Create new empty workflow.

install_requirements(path)

Install Python requirements from all algorithms of the workflow stored at the given path.

load(path)

Load Ikomia workflow from the given path.

prepare_runtime_env(workflow_path)

Install all algorithms needed to execute the workflow stored at the given path.

Details

ikomia.dataprocess.workflow.create(name: str = 'untitled')

Create new empty workflow. See also add_task(), connect_tasks() and run().

Parameters:

name (str) – workflow name.

Returns:

workflow instance

Return type:

Workflow

ikomia.dataprocess.workflow.install_requirements(path: str) bool

Install Python requirements from all algorithms of the workflow stored at the given path. Algorithms must be installed locally before calling this function.

Parameters:

path (str) – path to workflow definition file (.json)

Returns:

True if all installations succeeded else False

ikomia.dataprocess.workflow.load(path: str) Workflow

Load Ikomia workflow from the given path.

Parameters:

path (str)

Returns:

loaded workflow.

Return type:

Workflow

ikomia.dataprocess.workflow.prepare_runtime_env(workflow_path: str)

Install all algorithms needed to execute the workflow stored at the given path. If algorithms are not locally installed, the function will try to install them from Ikomia HUB (public and private if authenticated).

Parameters:

workflow_path (str) – path to workflow definition file (.json)