SDK Reference¶

Annotation¶

Working with labeling data of individual images. Annotation wraps all the labeling data for a given image: its Labels (geometrical objects) and Tags.

`Annotation`	Annotation for a single image.
`AnnotationJsonFields`	JSON fields for `Annotation`.
`Label`	Supervisely Label.
`LabelJsonFields`	JSON fields for `Label`.
`ObjClass`	Object class: name, geometry type (Rectangle, Polygon, etc.), color.
`ObjClassJsonFields`	Json fields for `ObjClass`
`ObjClassCollection`	Collection with `ObjClass` instances.
`Tag`	Tag attached to image or label; meta + value.
`TagJsonFields`	Json fields for `Tag`
`TagCollection`	Collection with `Tag` instances.
`TagMeta`	Tag metadata: name, value type (NONE, ANY_STRING, DATE, etc.), optional possible values.
`TagMetaJsonFields`	Json fields for `TagMeta`
`TagApplicableTo`	Defines Tag applicability only to images, objects or both.
`TagValueType`	Restricts Tag to have a certain value type.
`TagMetaCollection`	Collection with `TagMeta` instances.

API¶

Python wrappers to script your interactions with the Supervisely web instance. Instead of clicking around, you can write a script to request, via the API, a sequence of tasks (e.g. training a neural network and then running inference on a validation dataset).

`Api`	Main Supervisely API client (aggregates sub-APIs and handles auth, retries and common configuration).
`AgentApi`	API for working with agents.
`AnnotationApi`	API for working with image annotations.
`AppApi`	API for launching and managing Supervisely Apps (built on top of `TaskApi`).
`WorkflowSettings`	Used to customize the appearance and behavior of the workflow node.
`WorkflowMeta`	Used to customize the appearance of the workflow main and/or relation node.
`DatasetApi`	API for working with datasets.
`FileApi`	API for working with files in Team Files.
`GuidesApi`	API for working with Guides.
`StorageApi`	API for working with files and folders in Team Files and Cloud Storage.
`ImageAnnotationToolApi`	Control the Supervisely Image Annotation Tool UI via API.
`ImageAnnotationToolAction`	Action to set the figure as the current figure (selected figure) in the annotation tool.
`ImageApi`	API for working with images.
`ImageInfo`	NamedTuple with image information from Supervisely.
`BlobImageInfo`	Object with image parameters that describes image in blob file.
`ImportStorageApi`	Work with internal import storage in Supervisely.
`IssuesApi`	API for working with annotation issues and comments.
`LabelingJobApi`	API for working with labeling jobs.
`EntitiesCollectionApi`	API for working with entities collections.
`LabelingQueueApi`	API for working with labeling queues.
`ApiField`	Fields for API requests.
`ModuleApi`	Base class for entities that have a parent object in the system.
`ModuleApiBase`	Base class for all module APIs.
`CloneableModuleApi`	API for cloning entities.
`UpdateableModule`	API for updating entities.
`RemoveableModuleApi`	API for removing entities.
`RemoveableBulkModuleApi`	API for removing multiple entities.
`ModuleWithStatus`	API for getting and raising for status of an entity.
`ObjectClassApi`	API for working with ObjClass objects.
`PluginApi`	API for working with plugins.
`ProjectApi`	API for working with projects.
`ProjectClassApi`	API for working with classes in a project.
`RoleApi`	API for working with roles.
`TaskApi`	API for working with tasks.
`ActivityAction`	List of Team Actions to sort Team Activity.
`TeamApi`	API for working with teams.
`UserApi`	API for working with users.
`WorkspaceApi`	API for working with workspaces.

Video API¶

API for working with videos in Supervisely.

`VideoApi`	API for working with videos.
`VideoInfo`	NamedTuple with video information from Supervisely.
`VideoAnnotationAPI`	API for working with video annotations.
`VideoFigureApi`	API for working with `VideoFigure`.
`VideoFrameAPI`	API for working with `Frame`.
`VideoObjectApi`	API for working with video objects.
`VideoTagApi`	API for working with `VideoTag`.

Volume API¶

API for working with volumes in Supervisely.

`VolumeApi`	API for working with volumes.
`VolumeInfo`	NamedTuple with Volume information from Supervisely.
`VolumeAnnotationAPI`	API for working with volume annotations.
`VolumeFigureApi`	API for working with `VolumeFigure`.
`VolumeObjectApi`	API for working with `VolumeObject` in `VolumeAnnotation`.
`VolumeTagApi`	API for working with `VolumeTag`.

Pointcloud API¶

API for working with pointclouds in Supervisely.

`PointcloudApi`	API for working with point clouds.
`PointcloudInfo`	NamedTuple with pointcloud information from Supervisely.
`PointcloudEpisodeApi`	API for working with point cloud episodes.
`PointcloudAnnotationAPI`	API for working with PointcloudAnnotation.
`PointcloudEpisodeAnnotationAPI`	API for working with point cloud episode annotations.
`PointcloudFigureApi`	API for working with `PointcloudFigure`.
`PointcloudObjectApi`	API for working with point cloud objects.
`PointcloudTagApi`	API for working with `PointcloudTag`.

Neural Networks API¶

APIs for training, deploying models, running inference, and working with model metadata in Supervisely.

`DeployApi`	API for deploying models and controlling serving apps.
`NeuralNetworkApi`	API to interact with neural networks in Supervisely.
`TrainApi`	High-level API to start a training application on the Supervisely instance.
`ModelAPI`	Client for interacting with a deployed model (load settings/metadata, run inference).
`Prediction`	A single prediction result.
`PredictionSession`	Asynchronous inference session that yields `Prediction`.

Augmentation¶

Data augmentations to create more data variety for neural networks training.

`aug`	Augmentations for images and annotations

Collection¶

Key-indexed collections and helpers used across the SDK.

`DuplicateKeyError`	Raised when trying to add already existing key to `KeyIndexedCollection`.
`KeyObject`	Base class for objects that should implement the `key` method.
`KeyIndexedCollection`	Base class for collections that can be indexed by a key.
`MultiKeyIndexedCollection`	Collection that can be indexed by multiple keys.

Decorators¶

Small utility decorators (e.g. profiling helpers).

`profile`	Profile decorators for Supervisely.

Geometry¶

All the logic concerned with working with geometric objects - compute statistics like object area, transform (rotate, scale, shift), extract bounding boxes, compute intersections and more.

`AnyGeometry`	Placeholder geometry that accepts any shape type.
`Bitmap`	Binary or grayscale 2D mask with origin; used for segmentation.
`SkeletonizeMethod`	Specifies possible skeletonization methods of `Bitmap`.
`AlphaMask`	Bitmap mask with per-pixel alpha; used for semi-transparent segmentation.
`BitmapBase`	Base class for bitmap-like geometries (Bitmap, AlphaMask, MultichannelBitmap); mask + origin.
`Cuboid`	3D cuboid defined by eight corner points and six faces.
`CuboidFace`	A single face of a `Cuboid`.
`Cuboid3d`	3D cuboid with position, rotation, and dimensions (Vector3d).
`Vector3d`	A simple 3D vector (x, y, z) used by `Cuboid3d`.
`Mask3D`	3D volumetric mask (voxel data).
`Geometry`	Base class for all geometry classes.
`Point`	2D point at (row, col).
`Point3d`	3D point (x, y, z) in volume or point cloud space.
`PointLocation`	2D pixel coordinate (row, col); used for point positions and polygon vertices.
`Polygon`	Closed 2D polygon with exterior contour and optional interior holes.
`Polyline`	Open 2D polyline (sequence of connected line segments).
`Rectangle`	Axis-aligned 2D bounding box (top, left, bottom, right).
`ImageRotator`	Rotates images and geometry figures around image center by a given angle.
`VectorGeometry`	Base class for vector geometries (polygon, polyline) defined by exterior/interior contours.
`Node`	Single vertex in a keypoint graph: a labeled point (row, col) used as a building block for `GraphNodes` (e.g. pose keypoints, skeleton joints).
`GraphNodes`	Geometry representing a graph of labeled keypoints: a collection of `Node` objects for pose estimation, skeletons, or structured point sets (e.g. nose, left_eye, right_shoulder).

Imaging¶

Wrappers for working with images: IO, transformations, text rendering, color conversions.

`font`	Font utilities.
`color`	Color utilities.
`image`	Image I/O and basic image processing utilities.

IO¶

Low-level convenience IO wrappers that are used across the SDK.

`fs`	File system utilities for Supervisely.
`json`	JSON utilities for Supervisely.

Labeling Jobs¶

Helpers for working with labeling jobs and related utilities.

`utils`	utilities used for labeling jobs

Pointcloud¶

Core pointcloud-related helpers and utilities.

`pointcloud`	Functions for processing pointclouds

Pointcloud Annotation¶

Data model for pointcloud annotations (figures, objects, tags, episodes).

`PointcloudAnnotation`	Annotation for a single point cloud item in Supervisely format.
`PointcloudFigure`	Figure in point cloud annotation: 3D geometry (e.g. Cuboid3d) at a frame.
`PointcloudObject`	Object in point cloud annotation (obj_class + tags).
`PointcloudObjectCollection`	Collection with `PointcloudObject` instances.
`PointcloudTag`	Tag on point cloud annotation (meta, value).
`PointcloudTagCollection`	Collection with `PointcloudTag` instances.
`PointcloudEpisodeAnnotation`	Annotation for a point cloud episode (a sequence of frames) in Supervisely format.
`PointcloudEpisodeFrame`	Single frame in point cloud episode; holds PointcloudFigures at given index.
`PointcloudEpisodeFrameCollection`	Collection with `PointcloudEpisodeFrame` instances.
`PointcloudEpisodeTag`	Tag on point cloud episode or frame range (meta, value, frame_range).
`PointcloudEpisodeTagCollection`	Collection with `PointcloudEpisodeTag` instances.

Pointcloud Episodes¶

Helpers and data structures for pointcloud episodes.

`pointcloud_episodes`	Functions for processing pointcloud episodes

Project¶

Working with Supervisely projects on disk.

`Project`	A local Supervisely project directory that contains one or more datasets.
`ProjectType`	Enumerates supported Supervisely project types (images, videos, volumes, point clouds, episodes).
`ProjectMeta`	Project-level metadata: object classes, tag metas, project type and settings.
`ProjectMetaJsonFields`	JSON field names used in `ProjectMeta` serialization.
`Dataset`	A dataset directory inside a local Supervisely project.
`DatasetDict`	`Dataset` collection of `Project`.
`OpenMode`	Defines the mode of using the `Project` and `Dataset`.
`ItemPaths`	Paths to an item's data file and its annotation file within a local Supervisely dataset.
`ItemInfo`	Basic information about a dataset item and where it is stored on disk.
`VideoProject`	A local Supervisely project for video data.
`VideoDataset`	A dataset directory for video items inside a local Supervisely video project.
`VideoItemPaths`	Paths to a video's data file and its annotation file within a local Supervisely dataset.
`PointcloudProject`	A local Supervisely project for point cloud data.
`PointcloudDataset`	A dataset directory for point cloud items inside a local Supervisely point cloud project.
`VolumeProject`	A local Supervisely project for 3D volume data.
`VolumeDataset`	A dataset directory for 3D volume items inside a local Supervisely volume project.
`VersionInfo`	Object with parameters from Supervisely that describes the version of the project.
`DataVersion`	Class for managing project versions.

Task¶

Constants defining the directories where a plugin should expect the input and output data to be, plus helpers to stream progress data from a running plugin back to the web instance.

`Progress`	Progress reporter for long-running operations.

Training¶

High-level wrappers and helpers for building training applications (GUI, data prep, artifacts upload, benchmarking).

`TrainApp`	High-level wrapper for building Supervisely training applications.

User¶

User-related data structures and constants.

`UserRoleName`	Enumerates supported Supervisely user roles (e.g. admin, manager, annotator).

Video¶

Core video-related helpers and utilities.

`video`	Functions for processing videos

Video Annotation¶

Data model for video annotations (frames, objects, figures, tags).

`VideoAnnotation`	Annotation for a single video in Supervisely format.
`Frame`	Single frame in a video annotation; holds figures at a given index.
`FrameCollection`	Collection of Frame instances indexed by frame index.
`KeyIdMap`	Bidirectional mapping between UUID keys and integer IDs for video annotations.
`VideoFigure`	Figure in video annotation: geometry tied to a VideoObject at a frame index.
`VideoObject`	Tracked object across video frames; obj_class plus optional tags.
`VideoObjectCollection`	Collection of VideoObject instances indexed by key.
`VideoTag`	Tag applied to video or frame range (meta, value, frame_range).
`VideoTagCollection`	Collection with `VideoTag` instances.

Volume¶

Core volume-related helpers and utilities.

`volume`	Functions for processing volumes

Volume Annotation¶

Data model for volume annotations (objects, figures, tags, and slices).

`VolumeAnnotation`	Annotation for a 3D volume: objects, tags, planes, spatial figures.
`VolumeFigure`	Figure in volume annotation: geometry (Mask3D/ClosedSurfaceMesh) at a slice or spatial position.
`VolumeObject`	Object in volume annotation: obj_class, optional Mask3D geometry, tags.
`VolumeObjectCollection`	Collection with `VolumeObject` instances.
`VolumeTag`	Tag on a volume annotation (meta, value).
`VolumeTagCollection`	Collection with `VolumeTag` instances.
`Plane`	Plane in volume (sagittal, coronal, axial); collection of Slices.
`Slice`	Single slice in volume plane; holds VolumeFigures at given index.

Utility Functions¶

A collection of useful utility functions for common tasks in the Supervisely SDK.

`download_fast`	Download project in a fast mode.