TagMeta¶

class TagMeta(name, value_type, possible_values=None, color=None, sly_id=None, hotkey=None, applicable_to=None, applicable_classes=None, target_type=None)[source]¶

Bases: KeyObject, JsonSerializable

Tag metadata: name, value type (NONE, ANY_STRING, etc.), optional possible values. Immutable.

Parameters:

name : str¶: Tag name.
value_type : str¶: TagValueType: NONE, ANY_STRING, ANY_NUMBER, ONEOF_STRING.
possible_values : List[str], optional¶: Required for ONEOF_STRING; list of allowed values.
color : List[int, int, int], optional¶: RGB color [R, G, B]. Random if not provided.
sly_id : int, optional¶: Server-side tag meta ID.
hotkey : str, optional¶: Hotkey in annotation UI.
applicable_to : str, optional¶: TagApplicableTo: ALL, IMAGES_ONLY, OBJECTS_ONLY.
applicable_classes : List[str], optional¶: Restrict to specific class names.
target_type : str, optional¶: TagTargetType: ALL, FRAME_BASED, GLOBAL.

Raises:

ValueError – If value_type or color is invalid; ONEOF_STRING requires possible_values.

Usage Example:

import supervisely as sly

meta_dog = sly.TagMeta('dog', sly.TagValueType.NONE)
meta_cat = sly.TagMeta('cat', sly.TagValueType.ANY_STRING, applicable_to=sly.TagApplicableTo.OBJECTS_ONLY)
colors = ["brown", "white", "black"]
meta_coat = sly.TagMeta('coat color', sly.TagValueType.ONEOF_STRING, possible_values=colors, color=[255, 120, 0])

Methods

`add_possible_value`	Adds a new value to the list of possible values.
`clone`	Clone makes a copy of TagMeta with new fields, if fields are given, otherwise it will use original TagMeta fields.
`from_json`	Convert a json dict to TagMeta.
`get_header_ptable`
`get_row_ptable`
`is_compatible`	Check if the current TagMeta instance is compatible with another TagMeta instance.
`is_valid_value`	Checks value against object value type to make sure that value is valid.
`key`
`to_json`	Convert the TagMeta to a json dict.

Attributes

`applicable_classes`	Applicable classes.
`applicable_to`	Tag applicability to objects, images, or both.
`color`	`[R,G,B]` color.
`hotkey`	Hotkey for Tag in annotation tool UI.
`name`	Name.
`possible_values`	Possible values of object.
`sly_id`	Tag ID in Supervisely server.
`target_type`	Tag target type (scope) - entities, frames or both.
`value_type`	Value type.

classmethod from_json(data)[source]¶

Convert a json dict to TagMeta. Read more about Supervisely format.

Parameters:

data : dict¶: TagMeta in json format as a dict.

Returns:

TagMeta object

Return type:

TagMeta

Usage Example:

import supervisely as sly

data = {
    "name":"Color",
    "value_type":"oneof_string",
    "color":"#FF7800",
    "values":[
        "brown",
        "white",
        "black",
        "red",
        "blue",
        "yellow",
        "grey"
    ],
    "hotkey":"M",
    "applicable_type":"all",
    "classes":[
        "car",
        "bicycle"
    ]
}

meta_colors = sly.TagMeta.from_json(data)

classmethod get_header_ptable()[source]¶

add_possible_value(value)[source]¶

Adds a new value to the list of possible values.

Parameters:

value : str¶: New value that will be added to a list.

Raises:

ValueError – if object’s value type is not “oneof_string” or already exists in a list

Returns:

New instance of TagMeta object

Return type:

TagMeta

Usage Example:

import supervisely as sly

#In order to add possible values, you must first initialize a variable where all possible values will be stored if it doesnt exist already
colors = ["brown", "white", "black", "red", "chocolate", "gold", "grey"]
meta_coat_color = sly.TagMeta('coat color', sly.TagValueType.ONEOF_STRING, possible_values=colors, applicable_classes=["dog", "cat"])

print(meta_coat_color.possible_values)
# Output: ['brown', 'white', 'black', 'red', 'chocolate', 'gold', 'grey']

#Now we can add new possible value to our TagMeta
# Remember that TagMeta object is immutable, and we need to assign new instance of TagMeta to a new variable
meta_coat_color = meta_coat_color.add_possible_value("bald (no coat)")

print(meta_coat_color.possible_values)
# Output: ['brown', 'white', 'black', 'red', 'chocolate', 'gold', 'grey', 'bald (no coat)']

clone(name=None, value_type=None, possible_values=None, color=None, sly_id=None, hotkey=None, applicable_to=None, applicable_classes=None, target_type=None)[source]¶

Clone makes a copy of TagMeta with new fields, if fields are given, otherwise it will use original TagMeta fields.

Parameters:

name : str¶: Tag name.
value_type : str¶: Tag value type.
possible_values : List[str], optional¶: List of possible values.
color : List[int, int, int], optional¶: [R, G, B] color, generates random color by default.
sly_id : int, optional¶: Tag ID in Supervisely server.
hotkey : str, optional¶: Hotkey for Tag in annotation tool UI.
applicable_to : str, optional¶: Defines applicability of Tag only to images, objects or both.
applicable_classes : List[str], optional¶: Defines applicability of Tag only to certain classes.

Returns:

New instance of TagMeta object

Return type:

TagMeta

Usage Example:

import supervisely as sly

#Original TagMeta
meta_dog_breed = sly.TagMeta('breed', sly.TagValueType.NONE)

# TagMetas made of original TagMeta
# Remember that TagMeta class object is immutable, and we need to assign new instance of TagMeta to a new variable
A_breeds = ["Affenpinscher", "Afghan Hound", "Aidi", "Airedale Terrier", "Akbash Dog", "Akita"]
meta_A_breed = meta_dog_breed.clone(value_type=sly.TagValueType.ONEOF_STRING, possible_values=A_breeds, hotkey='A')

B_breeds = ["Basset Fauve de Bretagne", "Basset Hound", "Bavarian Mountain Hound", "Beagle", "Beagle-Harrier", "Bearded Collie"]
meta_B_breed = meta_A_breed.clone(possible_values=B_breeds, hotkey='B')

C_breeds = ["Cairn Terrier", "Canaan Dog", "Canadian Eskimo Dog", "Cane Corso", "Cardigan Welsh Corgi", "Carolina Dog"]
meta_C_breed = meta_B_breed.clone(possible_values=C_breeds, hotkey='C')

get_row_ptable()[source]¶

is_compatible(other)[source]¶: Check if the current TagMeta instance is compatible with another TagMeta instance.

is_valid_value(value)[source]¶

Checks value against object value type to make sure that value is valid.

Parameters:

value : str¶: Value to check.

Returns:

True if value is supported, otherwise False

Return type:

bool

Usage Example:

import supervisely as sly

# Initialize TagMeta
meta_dog = sly.TagMeta('dog', sly.TagValueType.ANY_STRING)

# Check what value type is in our Tagmeta
print(meta_dog.value_type)
# Output: 'any_string'

# Our TagMeta has 'any_string' value type, it means only 'string' values will work with it
# Let's check if value is valid for our TagMeta
meta_dog.is_valid_value('Woof!')            # True
meta_dog.is_valid_value(555)                # False

# TagMetas with 'any_number' value type are compatible with 'int' and 'float' values
meta_quantity = sly.TagMeta('quantity', sly.TagValueType.ANY_NUMBER)

meta_quantity.is_valid_value('new string value') # False
meta_quantity.is_valid_value(555)                # True
meta_quantity.is_valid_value(3.14159265359)      # True

to_json()[source]¶

Convert the TagMeta to a json dict. Read more about Supervisely format.

Returns:

Json format as a dict

Return type:

dict

Usage Example:

import supervisely as sly

colors = ["brown", "white", "black", "red", "blue", "yellow", "grey"]
meta_color = sly.TagMeta(
    'Color',
    sly.TagValueType.ONEOF_STRING,
    possible_values=colors,
    color=[255, 120, 0],
    hotkey="M",
    applicable_classes=["car", "bicycle"]
)


meta_color_json = meta_color.to_json()
print(meta_color_json)
# Output: {
#     "name":"Color",
#     "value_type":"oneof_string",
#     "color":"#FF7800",
#     "values":[
#         "brown",
#         "white",
#         "black",
#         "red",
#         "blue",
#         "yellow",
#         "grey"
#     ],
#     "hotkey":"M",
#     "applicable_type":"all",
#     "classes":[
#         "car",
#         "bicycle"
#     ]
# }

property applicable_classes¶

Applicable classes.

Returns:

List of applicable classes

Return type:

List[str]

Usage Example:

# Imagine we have 2 ObjClasses in our Project
class_car = sly.ObjClass(name='car', geometry_type='rectangle')
class_bicycle = sly.ObjClass(name='bicycle', geometry_type='rectangle')

# You can put a "string" with ObjClass name or use ObjClass.name
meta_vehicle = sly.TagMeta('vehicle', sly.TagValueType.NONE, applicable_classes=["car", class_bicycle.name])

print(meta_vehicle.applicable_classes)
# Output: ['car', 'bicycle']

property applicable_to¶

Tag applicability to objects, images, or both.

Returns:

Applicability

Return type:

str

Usage Example:

meta_dog = sly.TagMeta('dog', sly.TagValueType.NONE, applicable_to=IMAGES_ONLY)

print(meta_dog.applicable_to)
# Output: 'imagesOnly'

property color¶

[R,G,B] color.

Returns:

Color

Return type:

List[int, int, int]

Usage Example:

meta_dog = sly.TagMeta('dog', sly.TagValueType.NONE, color=[255,120,0])

print(meta_dog.color)
# Output: [255,120,0]

property hotkey¶

Hotkey for Tag in annotation tool UI.

Returns:

Hotkey

Return type:

str

Usage Example:

meta_dog = sly.TagMeta('dog', sly.TagValueType.NONE, hotkey='M')

print(meta_dog.hotkey)
# Output: 'M'

property name¶

Name.

Returns:

Name

Return type:

str

Usage Example:

meta_dog = sly.TagMeta('dog', sly.TagValueType.ANY_STRING)
print(meta_dog.name)
# Output: 'dog'

property possible_values¶

Possible values of object. This is a required field if object has “oneof_string” value type.

Raises:

ValueError – if list of possible values is not defined or TagMeta value_type is not “oneof_string”.

Returns:

List of possible values

Return type:

List[str]

Usage Example:

# List of possible values
coat_colors = ["brown", "white", "black", "red", "chocolate", "gold", "grey"]

# TagMeta
meta_coat_color = sly.TagMeta('coat color', sly.TagValueType.ONEOF_STRING, possible_values=coat_colors)

print(meta_coat_color.possible_values)
# Output: ['brown', 'white', 'black', 'red', 'chocolate', 'gold', 'grey']

# Note that this is a required field if object has "oneof_string" value type.
meta_coat_color = sly.TagMeta('coat color', sly.TagValueType.ONEOF_STRING)
# Output: ValueError: TagValueType is ONEOF_STRING. List of possible values have to be defined.

property sly_id¶

Tag ID in Supervisely server.

Returns:

Return type:

int

Usage Example:

meta_dog = sly.TagMeta('dog', sly.TagValueType.NONE, sly_id=38584)

print(meta_dog.sly_id)
# Output: 38584

property target_type¶

Tag target type (scope) - entities, frames or both.

Returns:

Target type

Return type:

str

Usage Example:

meta_dog = sly.TagMeta('dog', sly.TagValueType.NONE, target_type=TagTargetType.FRAME_BASED)

print(meta_dog.target_type)
# Output: 'framesOnly'

property value_type¶

Value type. See possible value types in TagValueType.

Returns:

Value type

Return type:

str

Usage Example:

meta_dog = sly.TagMeta('dog', sly.TagValueType.ANY_STRING)
meta_dog.value_type == sly.TagValueType.ANY_STRING # True

print(meta_dog.value_type)
# Output: 'any_string'