Properties

Property definitions and computed values.

`nimbusimage.properties.PropertyAccessor`

Access property definitions and values for a dataset.

`list()`

List all property definitions accessible to the user.

`get(property_id)`

Get a property definition by ID.

`create(name, shape='polygon', tags=None, image='properties/none:latest', worker_interface=None)`

Create a new property definition.

The new property is registered into all of this dataset's collections. Properties are only visible (via list() and get()) through collections that reference them, so an unregistered property would be invisible to every user, including its creator.

`get_or_create(name, shape='polygon', **kwargs)`

Get existing property by name+shape, or create it.

Only properties registered in a collection the user can read are findable, so the created property is registered into this dataset's collections (see create()).

`register(property_id)`

Add property to all collections for this dataset.

Fetches each unique collection, appends property_id if not present, and saves. Deduplicates by collection ID so shared collections are only updated once.

Returns:

Type	Description
`int`	The number of collections that reference the property
`int`	after registration. 0 means the dataset has no
`int`	collections and the property remains invisible.

`delete(property_id)`

Delete a property definition.

`get_values(annotation_id=None)`

Get property values.

Parameters:

Name	Type	Description	Default
`annotation_id`	`str \| None`	If provided, get values for this annotation only. Otherwise, get all values for the dataset.	`None`

`submit_values(property_id, values)`

Submit property values in bulk.

Transforms user-friendly format to backend wire format and auto-batches at 10K entries.

Parameters:

Name	Type	Description	Default
`property_id`	`str`	The property these values belong to.	required
`values`	`dict[str, dict]`	Dict mapping annotation_id to {key: value} dicts. Example: {"ann_1": {"Area": 100}, "ann_2": {"Area": 200}}	required

`delete_values(property_id)`

Delete all values for a property in this dataset.

`histogram(property_path, buckets=255)`

Get histogram for a property across all annotations.

`compute(property, worker_interface=None, scales=None)`

Run a property worker to compute values.

Submits a Docker worker job via POST /annotation_property/{id}/compute. The worker container receives the property definition and scales as JSON parameters.

Parameters:

Name	Type	Description	Default
`property`	`Property`	The Property definition. Must have a non-empty `image` field (the Docker image to run) and an `id` (must be saved to the server first via `create()` or `get_or_create()`).	required
`worker_interface`	`dict \| None`	Parameter values matching the worker's interface schema (from `client.get_worker_interface()`). If not provided, uses the property's own `workerInterface`.	`None`
`scales`	`dict \| None`	Scale metadata (pixel size, etc.). Passed through to the worker for unit-aware computations.	`None`

Returns:

Type	Description
`Job`	A Job object. Call `job.wait()` to block until completion.

Raises:

Type	Description
`ValueError`	If the property has no `id` or `image`, or if its `image` is an annotation worker (its role labels from `/worker_interface/available` have `isAnnotationWorker` but not `isPropertyWorker`) — annotation workers must run through `ds.annotations.compute` instead.