Supervisely
Last updated
Was this helpful?
Last updated
Was this helpful?
Easily import your volumes with annotations in the Supervisely format. The Supervisely json-based annotation format supports such figures: rectangle, line (polyline), polygon, point, bitmap (mask), graph (keypoints). It is a universal format that supports various types of annotations and is used in the Supervisely platform.
Supported volume formats: .nrrd, .dcm
With annotations: yes
Supported annotation format: .json.
Data structure: Information is provided below.
Example data: .
Both directory and archive are supported.
Recommended directory structure:
Root 📁 project_name folder named with the project name
📄 meta.json file
📄 key_id_map.json file (optional)
📁 dataset_name folders, each named with the dataset name and containing:
📁 volume folder, contains source volume files in , for example CTChest.nrrd
📁 ann - folder, with annotations for volumes. (named as volume + .json) for example CTChest.nrrd.json
📁 mask optional folder, created automatically while downloading project.
📁 folders, named according to volume (CTChest.nrrd), which contains an additional data files with geometries for annotation objects of class type Mask3D stored in , named with hex hash code of objects from key_id_map. For example: daff638a423a4bcfa34eb12e42243a87.nrrd
📁 interpolation ℹ️ optional folder, created automatically while downloading project.
📁 folders, named according to volume (CTChest.nrrd), which contains an additional data files in >), named with hex hash code of objects from key_id_map. For example: 24a56a26ed784e648d3dd6c5186b46ca.stl
ℹ️ - It is recommended to upload 3D objects as Mask3D and not to use STL. But if you already have a prepared STL file, all STL interpolations will be automatically converter to a Mask3D object during project upload.
Example:
annotation JSON file - /project_name/dataset_name/ann/CTChest.nrrd.json
volumeMeta - metadata for 3D reconstruction of volume
key - string - a unique identifier of given object represented as UUID.hex value (used in key_id_map.json to get the object ID)
tags - list of strings that will be interpreted as volume tags
objects - list of objects that may be present on the volume
spatialFigures - list of 3D figures may be present as the volume annotation
volumeMeta fields description:
intensity - {"min": int, "max": int} - intensity range. Depends on the device getting the data
windowWidth - float - Specify a linear conversion. Window Width contains the width of the window
windowCenter - float - Specify a linear conversion. Window Center contains the value that is the center of the window
channelsCount - float - channel count of your image data. Default: 1
Grayscale transformations to be applied to Pixel Data are defined by the equivalent of the Modality LUT and Rescale Intercept, Value of Interest Attributes, Photometric Interpretation and the equivalent of the Presentation LUT.
units = m*SV + b
rescaleSlope - float - m in the equation specified by Rescale Intercept
rescaleIntercept - float - The value "b" in the relationship between stored values (SV) in Pixel Data and the output units specified in Rescale Type.
objects fields description:
key - string - a unique identifier of given object represented as UUID.hex value (used in key_id_map.json to get the object ID)
classTitle - string - the title of a class. It's used to identify the class shape from the meta.json file
tags - list of strings that will be interpreted as object tags
labelerLogin - string - the name of the user that added this figure to the project
updatedAt - string - the date and time when the object was updated (ISO 8601 format)
createdAt - string - the date and time when the object was updated (ISO 8601 format)
planes fields description:
normal - dict with x, y, z as keys and 0/1 as values - normal is direction by axis, chosen according to plane name
slices - list of slices on the plane. Each list contains index and may contain figures.
slices fields description:
index - int value of slice index
spatialFigures fields description
key - string - unique key for a given figure (used in key_id_map.json)
objectKey - string - unique key to link figure to object (used in key_id_map.json)
geometryType - mask_3d or other 3D geometry-class shape
geometry - geometry of the object
mask folderThese files contain geometry for 3D annotation objects, every file name must be the same as figure key to which it belongs.
Example:
/project_name/dataset_name/mask/CTChest.nrrd/daff638a423a4bcfa34eb12e42243a87.nrrd connected with spatial figure "key": "daff638a423a4bcfa34eb12e42243a87"
/project_name/key_id_map.json file is optional. It is created when annotating the volume inside Supervisely interface and sets the correspondence between the unique identifiers of the object and the volume on which the figure is located. If you annotate manually, you do not need to create this file. This will not affect the work being done.
JSON file format of key_id_map.json:
objects - dictionary, where the key is a unique string, generated inside Supervisely environment to set mapping of current object in annotation, and value is unique integer ID related to the current object
figures - dictionary, where the key is a unique string, generated inside Supervisely environment to set mapping of object on volume in annotation, and value is unique integer ID related to the current volume
videos - dictionary, where the key is unique string, generated inside Supervisely environment to set mapping of volumes in annotation, and value is a unique integer ID related to the current volume
tags - dictionary, where the keys are unique strings, generated inside Supervisely environment to set mapping of tag on current volume in annotation, and value is a unique integer ID related to the current tag
Value - returned by server integer identifier while uploading object / figure / volume / tag.
planes - a list of figures that defined in these planes:
ACS - string - "RAS" or "LPS" - name of type of i.e. RAS means is Right-Anterior-Superior
dimensionsIJK - dict {"x": int, "y": int, "z": int} - dimensions of volume described as vector in
IJK2WorldMatrix - matrix to transform coordinates from IJK to world (cartesian). See
name - string - the name of the plane, where the figures are placed. Can be
figures - list of figures placed on a slice. It can be or .
This list contains 3D objects of type
Definitions for its fields can be found
Key - generated by . The unique string. All key and ID values should be unique inside single project and can not be shared between entities.
