Project Management

`get_info`

Retrieves project information based on the project key, such as the project type, asset groups, tags, and project's statistics.

Arguments

None

Return

Project object containing the relevant project information with the following structure:

Project(
    id='proj_9004a21df7b040ace4674c4879603fe8',
    name='keypoints',
    workspace_id='ws_1c8aab980f174b0296c7e35e88665b13',
    type='ObjectDetection',
    create_date=1701927649302,
    localization='MULTI',
    tags=['cat faces'],
    groups=['main', 'cats'],
    statistic=Statistic(
        tags_count=[TagsCountItem(name='cat faces', count=0)],
        total_assets=28,
        annotated_assets=0,
        total_annotations=0
    )
)

Attribute	Type	Description
`id`	`str`	Unique ID of the project.
`name`	`str`	Name of the project. You can modify this by calling the function `project.update({"name": "YOUR_NEW_PROJECT_NAME"})`, or you can directly modify this on Nexus by clicking on the Settings tab on the left sidebar of the Project Dashboard.
`workspace_id`	`str`	Unique ID of the current workspace that this project is in.
`type`	`str`	Type of project. This can be one of `ObjectDetection`, `InstanceSegmentation`, `Classification`, or `Keypoint` depending on the type selected during project creation on Nexus.
`create_date`	`int`	UNIX timestamp of project creation date.
`localization`	`str`	Region(s) for data localization. Defaults to `MULTI` for multi-region.
`tags`	`list[str]`	List of tag names in the project.
`groups`	`list[str`	List of asset group names in the project.
`statistic`	`Statistic` object	Contains project statistics for the following categories: - `tags_count`: List of `TagCountItem` objects representing tag counts for each tag in the project. - `total_assets`: Total number of assets in the project. - `annotated_assets`: Total number of annotated assets in the project. - `total_annotations`: Total number of annotations in the project.

Examples

Retrieve project information from the project with project key 9004a21df7b040ace4674c4879603fe8

from datature.nexus import Client

SECRET_KEY = "01048a3572a00dec13a6ca204fa78c0c4c216475c2ca80cb618114d26783d5a2"
PROJECT_ID = "proj_9004a21df7b040ace4674c4879603fe8"
client = Client(SECRET_KEY)
project = client.get_project(PROJECT_ID)

project.get_info()

`update`

Updates the metadata of the project, such as the name of the project.

Arguments

Name	Type	Description
`project`	`dict \|ProjectMetadata`	The metadata of the project.

Return

Project object containing the relevant project information with the following structure:

Project(
    id='proj_9004a21df7b040ace4674c4879603fe8',
    name='My Cool Project',
    workspace_id='ws_1c8aab980f174b0296c7e35e88665b13',
    type='ObjectDetection',
    create_date=1701927649302,
    localization='MULTI',
    tags=['cat faces'],
    groups=['main', 'cats'],
    statistic=Statistic(
        tags_count=[TagsCountItem(name='cat faces', count=0)],
        total_assets=28,
        annotated_assets=0,
        total_annotations=0
    )
)

Attribute	Type	Description
`id`	`str`	Unique ID of the project.
`name`	`str`	Name of the project. You can modify this by calling the function `project.update({"name": "YOUR_NEW_PROJECT_NAME"})`, or you can directly modify this on Nexus by clicking on the Settings tab on the left sidebar of the Project Dashboard.
`workspace_id`	`str`	Unique ID of the current workspace that this project is in.
`type`	`str`	Type of project. This can be one of `ObjectDetection`, `InstanceSegmentation`, `Classification`, or `Keypoint` depending on the type selected during project creation on Nexus.
`create_date`	`int`	UNIX timestamp of project creation date.
`localization`	`str`	Region(s) for data localization. Defaults to `MULTI` for multi-region.
`tags`	`list[str]`	List of tag names in the project.
`groups`	`list[str`	List of asset group names in the project.
`statistic`	`Statistic` object	Contains project statistics for the following categories: - `tags_count`: List of `TagCountItem` objects representing tag counts for each tag in the project. - `total_assets`: Total number of assets in the project. - `annotated_assets`: Total number of annotated assets in the project. - `total_annotations`: Total number of annotations in the project.

Examples

Updating the name of the project 9004a21df7b040ace4674c4879603fe8 to My Cool Project by providing the metadata as a dictionary.

from datature.nexus import Client

SECRET_KEY = "01048a3572a00dec13a6ca204fa78c0c4c216475c2ca80cb618114d26783d5a2"
PROJECT_ID = "proj_9004a21df7b040ace4674c4879603fe8"
client = Client(SECRET_KEY)
project = client.get_project(PROJECT_ID)

project.update({"name":"My Cool Project"})

Updating the name of the project 9004a21df7b040ace4674c4879603fe8 to My Cool Project by providing the metadata as a ProjectMetadata object.

from datature.nexus import Client

SECRET_KEY = "01048a3572a00dec13a6ca204fa78c0c4c216475c2ca80cb618114d26783d5a2"
PROJECT_ID = "proj_9004a21df7b040ace4674c4879603fe8"
client = Client(SECRET_KEY)
project = client.get_project(PROJECT_ID)

project.update(ApiTypes.ProjectMetadata(name="My Cool Project"))

`list_insights`

Retrieves project insight and metrics of the completed training runs.

Arguments

None

Return

List of ProjectInsight objects containing information and metrics on completed training runs.

[ProjectInsight(
    flow_title='Test workflow', 
    run_id='run_4a5d406d-464d-470c-bd7d-e92456621ad3', 
    dataset=InsightDataset(
        data_type='Rectangle', 
        num_classes=1, 
        average_annotations=5.19, 
        total_assets=500, 
        settings=DatasetSettings(
            split_ratio=0.3, 
            shuffle=True, 
            seed=0, 
            using_sliding_window=False
        )
    ), 
    model=InsightModel(
        name='fasterrcnn-inceptionv2-1024x1024', 
        batch_size=2, 
        training_steps=5000, 
        max_detection_per_class=100, 
        solver='momentum', 
        learning_rate=0.04, 
        momentum=0.9
    ), 
    checkpoint=RunCheckpoint(
        strategy='STRAT_ALWAYS_SAVE_LATEST', 
        evaluation_interval=250, 
        metric=None
    ), 
    artifact=InsightArtifact(
        id='artifact_65ae274540259e2a07533532', 
        is_training=False, 
        step=5000, 
        metric=ArtifactMetric(
            total_loss=0.32356, 
            classification_loss=0.012036, 
            localization_loss=0.010706, 
            regularization_loss=0.0
        )
    ),
    create_date=1705912133684
)]

Name	Type	Description
`flow_title`	`str`	Name of the workflow.
`run_id`	`str`	ID of the training run.
`dataset`	`InsightDataset` object	Information and statistics on the training dataset.
`model`	`InsightModel` object	Information on model architecture and hyperparameters.
`checkpoint`	`RunCheckpoint` object	Information on training setup.
`artifact`	`InsightArtifact` object	Information and metrics on saved artifact.
`create_date`	`int`	UNIX timestamp of the project creation date.

Examples

View the training insight of all training runs:

from datature.nexus import Client

SECRET_KEY = "01048a3572a00dec13a6ca204fa78c0c4c216475c2ca80cb618114d26783d5a2"
PROJECT_ID = "proj_9004a21df7b040ace4674c4879603fe8"
client = Client(SECRET_KEY)
project = client.get_project(PROJECT_ID)

project.list_insights()

View the training insight of all training runs with model architecture EfficientDet D1 640x640:

from datature.nexus import Client

SECRET_KEY = "01048a3572a00dec13a6ca204fa78c0c4c216475c2ca80cb618114d26783d5a2"
PROJECT_ID = "proj_9004a21df7b040ace4674c4879603fe8"
client = Client(SECRET_KEY)
project = client.get_project(PROJECT_ID)

insights = project.list_insights()
filtered = [
    run for run in insights if run.model.name == "efficientdet-d1-640x640"
]

`list_users`

Retrieves all users in the project, this includes Project Owners, Collaborators, and Datature Experts.

Arguments

None

Return

List of ProjectUser objects containing details of users who have access to the project, with the following structure:

[ProjectUser(
    id='user_6323fea23e292439f31c58cd',
    access_type='Owner',
    email='[email protected]',
    nickname='raighne',
    picture='https://s.gravatar.com/avatar/avatars%2Fra.png'
)]

Name	Type	Description
`id`	`str`	Unique ID of the user.
`access_type`	`str`	The role access of the user, this is one of `Owner`, `Collaborator`, `Labeller`, or `Expert` (for Datature engineering consultants)
`email`	`str`	Email that the user signed up with.
`nickname`	`str`	Nickname of the user.
`picture`	`str`	URL to the user's profile picture.

Examples

List all users in the project with project key 9004a21df7b040ace4674c4879603fe8.

from datature.nexus import Client

SECRET_KEY = "01048a3572a00dec13a6ca204fa78c0c4c216475c2ca80cb618114d26783d5a2"
PROJECT_ID = "proj_9004a21df7b040ace4674c4879603fe8"
client = Client(SECRET_KEY)
project = client.get_project(PROJECT_ID)

project.list_users()