AppApi¶

Bases: TaskApi

API for launching and managing Supervisely Apps (built on top of TaskApi).

Parameters:

api¶: Api object to use for API connection.

Methods

`convert_info_to_json`	Convert information about an entity to a dictionary.
`create_task_detached`	create_task_detached
`deploy_model`	deploy_model
`download_git_archive`	download_git_archive
`download_git_file`	Download a file from app repository.
`download_import_file`	download_import_file
`find_module_id_by_app_name`
`get_context`	Get context information by Application Task ID.
`get_ecosystem_module_id`	Returns ecosystem module id by slug.
`get_ecosystem_module_info`	Returns ModuleInfo object by module id and version.
`get_field`	get_field
`get_fields`	get_fields
`get_import_files_list`	get_import_files_list
`get_info`	get_info
`get_info_by_id`	Get application info by numeric ID.
`get_list`	Get list of applications for the specified team.
`get_list_all_pages`	Get list of all or limited quantity entities from the Supervisely server.
`get_list_all_pages_generator`	This generator function retrieves a list of all or a limited quantity of entities from the Supervisely server, yielding batches of entities as they are retrieved
`get_list_ecosystem_modules`
`get_list_idx_page_async`	Get the list of items for a given page number.
`get_list_page_generator_async`	Yields list of images in dataset asynchronously page by page.
`get_logs`	Returns the log entries for the given task ID in JSON format.
`get_session_token`	Get session token for the app with specified slug.
`get_sessions`	Get list of sessions (tasks) for the specified team and module.
`get_status`
`get_training_metrics`	get_training_metrics
`get_url`	get_url
`info_sequence`	Sequence of fields that are returned by the API to represent AppInfo.
`info_tuple_name`	Name of the tuple that represents AppInfo.
`initialize`	initialize
`is_ready`	Check if the task is ready.
`is_ready_for_api_calls`	Checks if app is ready for API calls.
`is_running`	Check if the task is running.
`raise_for_status`	Raise error if Application Task status is ERROR.
`run_dtl`	run_dtl
`run_inference`	run_inference
`run_train`	run_train
`send_request`	send_request
`set_field`	set_field
`set_fields`	set_fields
`set_fields_from_dict`	set_fields_from_dict
`set_output_archive`	set_output_archive
`set_output_directory`	set_output_directory
`set_output_error`	Set custom error message to the task output.
`set_output_experiment`	Sets output for the task with experiment info.
`set_output_file_download`	set_output_file_download
`set_output_project`	set_output_project
`set_output_report`	set_output_report
`set_output_text`	Set custom text message to the task output.
`start`	Start a new application session (task).
`stop`	stop
`submit_logs`	submit_logs
`update_meta`	Update given task metadata :param id: int - task id :param data: dict - meta data to update
`update_status`	Sets the specified status for the task.
`upload_files`	upload_files
`wait`	wait
`wait_until_ready_for_api_calls`	Waits until app is ready for API calls.

Attributes

`MAX_WAIT_ATTEMPTS`	Maximum number of attempts that will be made to wait for a certain condition to be met.
`WAIT_ATTEMPT_TIMEOUT_SEC`	Number of seconds for intervals between attempts.

InfoType¶: alias of AppInfo

class PluginTaskType(*values)¶

Bases: StrEnum

Task type for plugin-based tasks (train/inference/smarttool/etc.).

classmethod has_value(value)¶

Check if a value is in the enum.

Parameters:

value : str¶: Value to check.

Returns:

True if value is in the enum, False otherwise.

Return type:

bool

classmethod values()¶: Get all values of the enum.

class RestartPolicy(*values)¶

Bases: StrEnum

Task restart policy used for app deployments.

classmethod has_value(value)¶

Check if a value is in the enum.

Parameters:

value : str¶: Value to check.

Returns:

True if value is in the enum, False otherwise.

Return type:

bool

classmethod values()¶: Get all values of the enum.

class Status(*values)¶

Bases: StrEnum

Task lifecycle status values returned by the API.

classmethod has_value(value)¶

Check if a value is in the enum.

Parameters:

value : str¶: Value to check.

Returns:

True if value is in the enum, False otherwise.

Return type:

bool

classmethod values()¶: Get all values of the enum.

CONSUMED = 'consumed'¶: Application is consumed by an agent

DEPLOYED = 'deployed'¶: Only for Plugins

ERROR = 'error'¶: Application has finished with an error

FINISHED = 'finished'¶: Application has finished successfully

QUEUED = 'queued'¶: Application is queued for execution

STARTED = 'started'¶: Application has been started

STOPPED = 'stopped'¶: Application has been stopped

TERMINATING = 'terminating'¶: Application is being terminated

class Workflow(api)[source]¶

Bases: object

Workflow functionality for connecting project states and task application sessions.

Parameters:

api¶: Api object to use for API connection.

add_input_dataset(dataset, task_id=None, meta=None)[source]¶

Add input type “dataset” to the workflow node. This type is used to show that the application has used the specified dataset. Customization of the dataset node is not supported and will be ignored. You can only customize the main node with this method.

Parameters:

dataset¶: Dataset ID or DatasetInfo object.
task_id : Optional[int]¶: Task ID. If not specified, the task ID will be determined automatically.
meta=None¶: Additional data for node customization.

Returns:

Response from the API.

Return type:

dict

add_input_file(file, model_weight=False, task_id=None, meta=None)[source]¶

Add input type “file” to the workflow node. This type is used to show that the application has used the specified file.

Parameters:

file¶: File ID, FileInfo object or file path in team Files.
model_weight : bool¶: Flag to indicate if the file is a model weight.
task_id : Optional[int]¶: Task ID. If not specified, the task ID will be determined automatically.
meta=None¶: Additional data for node customization.

Returns:

Response from the API.

Return type:

dict

add_input_folder(path, task_id=None, meta=None)[source]¶

Add input type “folder” to the workflow node. Path to the folder is a path in Team Files. This type is used to show that the application has used files from the specified folder.

Parameters:

path : str¶: Path to the folder in Team Files.
task_id : Optional[int]¶: Task ID. If not specified, the task ID will be determined automatically.
meta=None¶: Additional data for node customization.

Returns:

Response from the API.

Return type:

dict

add_input_job(id, task_id=None, meta=None)[source]¶

Add input type “job” to the workflow node. Job is a Labeling Job. This type indicates that the application has utilized a labeling job during its operation.

Parameters:

id : int¶: Labeling Job ID.
task_id : Optional[int]¶: Task ID. If not specified, the task ID will be determined automatically.
meta=None¶: Additional data for node customization.

Returns:

Response from the API.

Return type:

dict

add_input_project(project=None, version_id=None, version_num=None, task_id=None, meta=None)[source]¶

Add input type “project” to the workflow node. The project version can be specified to indicate that the project version was used especially for this task. Arguments project and version_id are mutually exclusive. If both are specified, version_id will be used. Argument version_num can only be used in conjunction with the project. This type is used to show that the application has used the specified project. Customization of the project node is not supported and will be ignored. You can only customize the main node with this method.

Parameters:

project=None¶: Project ID or ProjectInfo object.
version_id : Optional[int]¶: Version ID of the project.
version_num : Optional[int]¶: Version number of the project. This argument can only be used in conjunction with the project.
task_id : Optional[int]¶: Task ID. If not specified, the task ID will be determined automatically.
meta=None¶: Additional data for node customization.

Returns:

Response from the API.

Return type:

dict

add_input_task(input_task_id, task_id=None, meta=None)[source]¶

Add input type “task” to the workflow node. This type usually indicates that the one application has used another application for its work.

Parameters:

input_task_id : int¶: Task ID that is used as input.
task_id : Optional[int]¶: Task ID of the node. If not specified, the task ID will be determined automatically.
meta=None¶: Additional data for node customization.

Returns:

Response from the API.

Return type:

dict

add_output_app(task_id=None, meta=None)[source]¶

Add output type “app_session” to the workflow node. This type is used to show that the application has an offline session in which you can find the result of its work.

Parameters:

task_id : Optional[int]¶: App Task ID. If not specified, the task ID will be determined automatically.
meta=None¶: Additional data for node customization.

Returns:

Response from the API.

Return type:

dict

add_output_dataset(dataset, task_id=None, meta=None)[source]¶

Add output type “dataset” to the workflow node. This type is used to show that the application has created a dataset with the result of its work. Customization of the dataset node is not supported and will be ignored. You can only customize the main node with this method.

Parameters:

dataset¶: Dataset ID or DatasetInfo object.
task_id : Optional[int]¶: Task ID. If not specified, the task ID will be determined automatically.
meta=None¶: Additional data for node customization.

Returns:

Response from the API.

Return type:

dict

add_output_file(file, model_weight=False, task_id=None, meta=None)[source]¶

Add output type “file” to the workflow node. This type is used to show that the application has created a file with the result of its work.

Parameters:

file¶: File ID or FileInfo object.
model_weight : bool¶: Flag to indicate if the file is a model weight.
task_id : Optional[int]¶: Task ID. If not specified, the task ID will be determined automatically.
meta=None¶: Additional data for node customization.

Returns:

Response from the API.

Return type:

dict

add_output_folder(path, task_id=None, meta=None)[source]¶

Add output type “folder” to the workflow node. Path to the folder is a path in Team Files. This type is used to show that the application has created a folder with the result files of its work.

Parameters:

path : str¶: Path to the folder.
task_id : Optional[int]¶: Task ID. If not specified, the task ID will be determined automatically.
meta=None¶: Additional data for node customization.

Returns:

Response from the API.

Return type:

dict

add_output_job(id, task_id=None, meta=None)[source]¶

Add output type “job” to the workflow node. Job is a Labeling Job. This type is used to show that the application has created a labeling job with the result of its work.

Parameters:

id : int¶: Labeling Job ID.
task_id : Optional[int]¶: Task ID. If not specified, the task ID will be determined automatically.
meta=None¶: Additional data for node customization.

Returns:

Response from the API.

Return type:

dict

add_output_project(project, version_id=None, task_id=None, meta=None)[source]¶

Add output type “project” to the workflow node. The project version can be specified with “version” argument to indicate that the project version was created especially as result of this task. This type is used to show that the application has created a project with the result of its work. Customization of the project node is not supported and will be ignored. You can only customize the main node with this method.

Parameters:

project¶: Project ID or ProjectInfo object.
version_id : Optional[int]¶: Version ID of the project.
task_id : Optional[int]¶: Task ID. If not specified, the task ID will be determined automatically.
meta=None¶: Additional data for node customization.

Returns:

Response from the API.

Return type:

dict

add_output_task(output_task_id, task_id=None, meta=None)[source]¶

Add output type “task” to the workflow node. This type is used to show that the application has created a task with the result of its work.

Parameters:

output_task_id : int¶: Created task ID.
task_id : Optional[int]¶: Task ID of the node. If not specified, the task ID will be determined automatically.
meta=None¶: Additional data for node customization.

Returns:

Response from the API.

Return type:

dict

check_instance_compatibility()[source]¶

Decorator to check instance compatibility with workflow features. If the instance is not compatible, the function will not be executed.

Parameters:

min_instance_version : Optional[str]: Determine the minimum instance version that accepts the workflow method. If not specified, the minimum version will be “6.9.31”.

disable()[source]¶: Disable the workflow functionality.

enable()[source]¶: Enable the workflow functionality.

is_enabled()[source]¶: Check if the workflow functionality is enabled.

classmethod convert_info_to_json(info)¶

Convert information about an entity to a dictionary.

Parameters:

info : NamedTuple¶: Information about the entity.

Returns:

Dictionary with information about the entity.

Return type:

dict

static info_sequence()[source]¶: Sequence of fields that are returned by the API to represent AppInfo.

static info_tuple_name()[source]¶: Name of the tuple that represents AppInfo.

create_task_detached(workspace_id, task_type=None)[source]¶

deploy_model(agent_id, model_id)[source]¶

download_git_archive(ecosystem_item_id, app_id, version, save_path, log_progress=True, ext_logger=None)[source]¶

download_git_file(module_id, save_path, app_id=None, version=None, file_path=None, file_key=None, log_progress=True, ext_logger=None)[source]¶

Download a file from app repository. File should be added in the app config under files key.

Parameters:

module_id : int¶: ID of the module
save_path : str¶: Path to save the file
app_id : int¶: ID of the app
version : str¶: Version of the app
file_path : str¶: Path to the file in the app github repository
file_key : str¶: Key of the file in the app github repository
log_progress : bool¶: If True, will log the progress of the download
ext_logger : Logger¶: Logger to use for logging

Returns:

None

Return type:

None

download_import_file(id, file_path, save_path)[source]¶

get_context(id)¶

Get context information by Application Task ID.

Parameters:

id : int¶: Task ID in Supervisely.

Returns:

Context information in dict format

Return type:

dict

Usage Example:

import os
from dotenv import load_dotenv

import supervisely as sly

# Load secrets and create API object from .env file (recommended)
# Learn more here: https://developer.supervisely.com/getting-started/basics-of-authentication
if sly.is_development():
    load_dotenv(os.path.expanduser("~/supervisely.env"))

api = sly.Api.from_env()

task_id = 121230
context = api.task.get_context(task_id)
print(context)
# Output: {
#     "team": {
#         "id": 16087,
#         "name": "alexxx"
#     },
#     "workspace": {
#         "id": 23821,
#         "name": "my_super_workspace"
#     }
# }

get_ecosystem_module_id(slug)[source]¶

Returns ecosystem module id by slug. E.g. slug = “supervisely-ecosystem/export-to-supervisely-format”. Slug can be obtained from the application URL in browser.

Parameters:

slug : str¶: module slug, starts with “supervisely-ecosystem/”

Returns:

ID of the module

Return type:

int

Raises:

KeyError – if module with given slug not found
KeyError – if there are multiple modules with the same slug

Usage Example:

import os
from dotenv import load_dotenv

import supervisely as sly

# Load secrets and create API object from .env file (recommended)
# Learn more here: https://developer.supervisely.com/getting-started/basics-of-authentication
load_dotenv(os.path.expanduser("~/supervisely.env"))
api = sly.Api.from_env()

slug = "supervisely-ecosystem/export-to-supervisely-format"
module_id = api.app.get_ecosystem_module_id(slug)
print(f"Module {slug} has id {module_id}")
# Module supervisely-ecosystem/export-to-supervisely-format has id 81

get_ecosystem_module_info(module_id=None, version=None, slug=None)[source]¶

Returns ModuleInfo object by module id and version.

Parameters:

module_id : int¶: ID of the module
version : Optional[str]¶: version of the module, e.g. “v1.0.0”
slug : Optional[str]¶: slug of the module, e.g. “supervisely-ecosystem/export-to-supervisely-format”

Raises:

ValueError – if both module_id and slug are None
ValueError – if both module_id and slug are provided

Returns:

ModuleInfo object

Return type:

ModuleInfo

Usage Example:

import os
from dotenv import load_dotenv

import supervisely as sly

# Load secrets and create API object from .env file (recommended)
# Learn more here: https://developer.supervisely.com/getting-started/basics-of-authentication
load_dotenv(os.path.expanduser("~/supervisely.env"))
api = sly.Api.from_env()

module_id = 81
module_info = api.app.get_ecosystem_module_info(module_id)

get_field(task_id, field)¶

get_fields(task_id, fields)¶

get_import_files_list(id)[source]¶

get_info(module_id, version=None)[source]¶

get_info_by_id(id)[source]¶

Get application info by numeric ID.

Parameters:

id : int¶: Application ID.

Returns:

Application info by numeric ID.

Return type:

AppInfo

get_list(team_id, filter=None, context=None, repository_key=None, show_disabled=False, integrated_into=None, session_tags=None, only_running=False, with_shared=True, force_all_sessions=True)[source]¶

Get list of applications for the specified team.

Parameters:

team_id : int¶

Team ID.

filter : Optional[List[dict]]¶

List of filters.

context : Optional[List[str]]¶

List of application contexts.

repository_key : Optional[str]¶

Repository key.

show_disabled : bool¶

Show disabled applications.

integrated_into : Optional[List[str]]¶

Destination(s) of the application. Available values:

panel
files
standalone
data_commander
image_annotation_tool
video_annotation_tool
dicom_annotation_tool
pointcloud_annotation_tool

session_tags : Optional[List[str]]¶

list of session tags

only_running : bool¶

show only running applications (status is one of “queued”/”consumed”/”started”/”deployed”)

with_shared : bool¶

include shared applications

force_all_sessions : bool¶

force to get all sessions (tasks) for each application. Works only if only_running is False. Note that it can be a long operation.

Returns:

list of applications

Return type:

List[AppInfo]

Usage Example:

import os
from dotenv import load_dotenv

import supervisely as sly

# Load secrets and create API object from .env file (recommended)
# Learn more here: https://developer.supervisely.com/getting-started/basics-of-authentication
if sly.is_development():
    load_dotenv(os.path.expanduser("~/supervisely.env"))

api = sly.Api.from_env()

team_id = 447

# Get list of all applications (including all tasks in `tasks` field)
apps = api.app.get_list(team_id=team_id)

# Get list of all applications (only running tasks included in `tasks` field)
apps = api.app.get_list(team_id=team_id, force_all_sessions=False)

# Get list of only running applications
apps = api.app.get_list(team_id=team_id, only_running=True)

# Get list of applications with specific filters
filter = [{"field": "moduleId", "operator": "=", "value": 428}]
apps = api.app.get_list(team_id=team_id, filter=filter)

get_list_all_pages(method, data, progress_cb=None, convert_json_info_cb=None, limit=None, return_first_response=False)¶

Get list of all or limited quantity entities from the Supervisely server.

Parameters:

method : str¶: Request method name
data : dict¶: Dictionary with request body info
progress_cb : Progress, optional¶: Function for tracking download progress.
convert_json_info_cb : Callable, optional¶: Function for convert json info
limit : int, optional¶: Number of entity to retrieve
return_first_response : bool, optional¶: Specify if return first response

Returns:

List of entities.

Return type:

List[dict]

get_list_all_pages_generator(method, data, progress_cb=None, convert_json_info_cb=None, limit=None, return_first_response=False)¶

This generator function retrieves a list of all or a limited quantity of entities from the Supervisely server, yielding batches of entities as they are retrieved

Parameters:

method : str¶: Request method name
data : dict¶: Dictionary with request body info
progress_cb : Progress, optional¶: Function for tracking download progress.
convert_json_info_cb : Callable, optional¶: Function for convert json info
limit : int, optional¶: Number of entity to retrieve
return_first_response : bool, optional¶: Specify if return first response

async get_list_idx_page_async(method, data)¶

Get the list of items for a given page number. Page number is specified in the data dictionary.

Parameters:

method : str¶: Method to call for listing items.
data : dict¶: Data to pass to the API method.

Returns:

List of items.

Return type:

Tuple[int, List[NamedTuple]]

async get_list_page_generator_async(method, data, pages_count=None, semaphore=None)¶

Yields list of images in dataset asynchronously page by page.

Parameters:

method : str¶: Method to call for listing items.
data : dict¶: Data to pass to the API method.
pages_count : int, optional¶: Preferred number of pages to retrieve if used with a per_page limit. Will be automatically adjusted if the pagesCount differs from the requested number.
semaphore=None¶: Semaphore for limiting the number of simultaneous requests.

Returns:

List of images in dataset.

Return type:

AsyncGenerator[List[ImageInfo]]

Usage Example:

import os
from dotenv import load_dotenv

import supervisely as sly

# Load secrets and create API object from .env file (recommended)
# Learn more here: https://developer.supervisely.com/getting-started/basics-of-authentication
if sly.is_development():
    load_dotenv(os.path.expanduser("~/supervisely.env"))

api = sly.Api.from_env()

method = 'images.list'
data = {'datasetId': 123456}

loop = sly.utils.get_or_create_event_loop()
images = loop.run_until_complete(api.image.get_list_generator_async(method, data))

get_logs(task_id, since_time=None, limit=None)¶

Returns the log entries for the given task ID in JSON format.

Parameters:

task_id : int¶: Task ID in Supervisely.
since_time : Optional[str]¶: Optional ISO 8601 formatted timestamp to filter logs that were created after the specified time.
limit : Optional[int]¶: Optional integer to limit the number of log entries returned.

Returns:

List of log entries in JSON format.

Return type:

List[Dict]

Usage Example:

since_time = "2026-03-30T14:28:58.816Z"
limit = 100
task_id = 12345
logs = api.task.get_logs(task_id, since_time=since_time, limit=limit)
for log_entry in logs:
    print(log_entry) # {'level': 'info', 'message': "Added object class... 'timestamp': '2026-03-30T14:28:56.525Z', 'task_id': 58595}

get_session_token(slug)[source]¶

Get session token for the app with specified slug.

Parameters:

slug : str¶: Slug of the app, e.g. “supervisely-ecosystem/hello-world-app”.

Returns:

Session token for the app.

Return type:

str

get_sessions(team_id, module_id, show_disabled=False, session_name=None, statuses=None, with_shared=False)[source]¶

Get list of sessions (tasks) for the specified team and module.

Parameters:

team_id : int¶: team id
module_id : int¶: application module id
show_disabled : bool¶: show disabled applications
session_name : Optional[str]¶: session name to filter sessions
statuses=None¶: list of statuses to filter sessions
with_shared : bool¶: include shared application sessions

Returns:

List of sessions

Return type:

List[SessionInfo]

Usage Example:

import os
from dotenv import load_dotenv

import supervisely as sly

# Load secrets and create API object from .env file (recommended)
# Learn more here: https://developer.supervisely.com/getting-started/basics-of-authentication
if sly.is_development():
    load_dotenv(os.path.expanduser("~/supervisely.env"))

api = sly.Api.from_env()

team_id = 447
module_id = 428

# Get list of all sessions for the specified team and module ID
sessions = api.app.get_sessions(team_id, module_id)

# Get list of sessions with specific statuses
from supervisely.api.task_api import TaskApi

statuses = [TaskApi.Status.STARTED]
sessions = api.app.get_sessions(team_id, module_id, statuses=statuses)

get_training_metrics(task_id)[source]¶

get_url(task_id)[source]¶

initialize(task_id, template, data=None, state=None)[source]¶

is_ready(task_id)¶

Check if the task is ready.

Parameters:

task_id : int¶: Task ID in Supervisely.

Returns:

True if the task is ready, False otherwise.

Return type:

bool

is_ready_for_api_calls(task_id)[source]¶: Checks if app is ready for API calls. :param task_id: ID of the running task. :type task_id: int :returns: True if app is ready for API calls, False otherwise.

is_running(task_id)¶

Check if the task is running.

Parameters:

task_id : int¶: Task ID in Supervisely.

Returns:

True if the task is running, False otherwise.

Return type:

bool

raise_for_status(status)¶

Raise error if Application Task status is ERROR.

Parameters:

status : Status¶: Status object.

Returns:

None

Return type:

None

run_dtl(workspace_id, dtl_graph, agent_id=None)[source]¶

run_inference(agent_id, input_project_id, input_model_id, result_project_name, inference_config=None)[source]¶

run_train(agent_id, input_project_id, input_model_id, result_nn_name, train_config=None)[source]¶

send_request(task_id, method, data, context={}, skip_response=False, timeout=60, outside_request=True, retries=10, raise_error=False)¶

set_field(task_id, field, payload, append=False, recursive=False)¶

set_fields(task_id, fields)¶

set_fields_from_dict(task_id, d)¶

set_output_archive(task_id, file_id, file_name, file_url=None)¶

set_output_directory(task_id, file_id, directory_path)¶

set_output_error(task_id, title, description=None, show_logs=True)¶

Set custom error message to the task output.

Parameters:

task_id : int¶: Application task ID.
title : str¶: Error message to be displayed in the task output.
description : Optional[str]¶: Description to be displayed in the task output.
show_logs : Optional[bool], default True¶: If True, the link to the task logs will be displayed in the task output.

Returns:

Response JSON.

Return type:

Dict

Usage Example:

import os
from dotenv import load_dotenv

import supervisely as sly

# Load secrets and create API object from .env file (recommended)
# Learn more here: https://developer.supervisely.com/getting-started/basics-of-authentication
if sly.is_development():
    load_dotenv(os.path.expanduser("~/supervisely.env"))

api = sly.Api.from_env()

task_id = 12345
title = "Something went wrong"
description = "Please check the task logs"
show_logs = True
api.task.set_output_error(task_id, title, description, show_logs)

set_output_experiment(task_id, experiment_info)¶

Sets output for the task with experiment info.

Parameters:

task_id : int¶: Task ID in Supervisely.
experiment_info : dict¶: Experiment info from TrainApp. See class ExperimentInfo.

Returns:

Server response JSON.

Return type:

dict

Usage Example:

experiment_info = {
    "experiment_name": "247 Lemons RT-DETRv2-M",
    "framework_name": "RT-DETRv2",
    "model_name": "RT-DETRv2-M",
    "task_type": "object detection",
    "project_id": 76,
    "project_version": {"id": 222, "version": 4},
    "task_id": 247,
    "model_files": {"config": "model_config.yml"},
    "checkpoints": [
        "checkpoints/best.pth",
        "checkpoints/checkpoint0025.pth",
        "checkpoints/checkpoint0050.pth",
        "checkpoints/last.pth",
    ],
    "best_checkpoint": "best.pth",
    "export": {"ONNXRuntime": "export/best.onnx"},
    "app_state": "app_state.json",
    "model_meta": "model_meta.json",
    "train_val_split": "train_val_split.json",
    "train_size": 4,
    "val_size": 2,
    "train_collection_id": 530,
    "val_collection_id": 531,
    "hyperparameters": "hyperparameters.yaml",
    "artifacts_dir": "/experiments/76_Lemons/247_RT-DETRv2/",
    "datetime": "2025-01-22 18:13:43",
    "evaluation_report_id": 12961,
    "evaluation_report_link": "https://app.supervisely.com/model-benchmark?id=12961",
    "evaluation_metrics": {"mAP": 0.994, "AP50": 1.0, "AP75": 1.0},  # and other metrics
    "primary_metric": "mAP",
    "logs": {"type": "tensorboard", "link": "/experiments/76_Lemons/247_RT-DETRv2/logs/"},
}

set_output_file_download(task_id, file_id, file_name, file_url=None, download=True)¶

set_output_project(task_id, project_id, project_name=None, project_preview=None)¶

set_output_report(task_id, file_id, file_name, description='Report')¶

set_output_text(task_id, title, description=None, show_logs=False, zmdi_icon='zmdi-comment-alt-text', icon_color='#33c94c', background_color='#d9f7e4')¶

Set custom text message to the task output.

Parameters:

task_id : int¶: Application task ID.
title : str¶: Text message to be displayed in the task output.
description : Optional[str]¶: Description to be displayed in the task output.
show_logs : Optional[bool], default False¶: If True, the link to the task logs will be displayed in the task output.
zmdi_icon : Optional[str], default "zmdi-comment-alt-text"¶: Icon class name from Material Design Icons set (ZMDI).
icon_color : Optional[str], default "#33c94c" (nearest Duron Jolly Green)¶: Icon color in HEX format.
background_color : Optional[str], default "#d9f7e4" (Cosmic Latte)¶: Background color in HEX format.

Returns:

Response JSON.

Return type:

Dict

Usage Example:

import os
from dotenv import load_dotenv

import supervisely as sly

# Load secrets and create API object from .env file (recommended)
# Learn more here: https://developer.supervisely.com/getting-started/basics-of-authentication
if sly.is_development():
    load_dotenv(os.path.expanduser("~/supervisely.env"))

api = sly.Api.from_env()

task_id = 12345
title = "Task is finished"
api.task.set_output_text(task_id, title)

start(agent_id, app_id=None, workspace_id=None, description='', params=None, log_level='info', users_id=None, app_version=None, is_branch=False, task_name='run-from-python', restart_policy='never', proxy_keep_url=False, module_id=None, redirect_requests={}, kubernetes_settings=None, multi_user_session=False)[source]¶

Start a new application session (task).

Parameters:

agent_id : int¶: ID of the agent to run the task on. If set None - the task will be run on the any available agent.
app_id : Optional[int]¶: Deprecated. Use module_id instead.
workspace_id : Optional[int]¶: ID of the workspace to run the task in. If not specified, the default workspace will be used.
description : str¶: Task description which will be shown in UI.
params : Optional[dict]¶: Task parameters which will be passed to the application.
log_level : Literal["info", "debug", "warning", "error"]¶: Log level for the task. Default is “info”.
users_id : Optional[int]¶: User ID for which will be created an instance of the application.
app_version : Optional[str]¶: Application version e.g. “v1.0.0” or branch name e.g. “dev”.
is_branch : bool¶: If the application version is a branch name, set this parameter to True.
task_name : str¶: Task name which will be shown in UI. Default is “run-from-python”.
restart_policy : str¶: When the app should be restarted: never or if error occurred.
proxy_keep_url : bool¶: For internal usage only.
module_id : Optional[int]¶: Module ID. Can be obtained from the apps page in UI.
redirect_requests : dict¶: For internal usage only in Develop and Debug mode.
kubernetes_settings=None¶: Kubernetes settings for the task. If not specified, default settings will be used.
multi_user_session : bool, default is False¶: If True, the application session will be created as multi-user. In this case, multiple users will be able to connect to the same application session. All users will have separate application states. Available only for applications that support multi-user sessions.

Returns:

SessionInfo object with information about the started task.

Return type:

SessionInfo

Raises:

ValueError – If both app_id and module_id are not provided.
ValueError – If both app_id and module_id are provided.

stop(id)[source]¶

submit_logs(logs)¶

update_meta(id, data, agent_storage_folder=None, relative_app_dir=None)¶: Update given task metadata :param id: int - task id :param data: dict - meta data to update

update_status(task_id, status)¶

Sets the specified status for the task.

Parameters:

task_id : int¶: Task ID in Supervisely.
status¶: Task status to set.

Raises:

ValueError – If the status value is not allowed.

upload_files(task_id, abs_paths, names, progress_cb=None)[source]¶

wait(id, target_status, attempts=None, attempt_delay_sec=None)[source]¶

wait_until_ready_for_api_calls(task_id, attempts=10, attempt_delay_sec=10)[source]¶

Waits until app is ready for API calls.

Parameters:

task_id : int¶: ID of the running task.
attempts : int¶: Number of attempts to check if app is ready for API calls.
attempt_delay_sec : int¶: Delay between attempts in seconds.

Returns:

True if app is ready for API calls, False otherwise.