CreateLabelingJob - Amazon SageMaker Service


Creates a job that uses workers to label the data objects in your input dataset. You can use the labeled data to train machine learning models.

You can select your workforce from one of three providers:

  • A private workforce that you create. It can include employees, contractors, and outside experts. Use a private workforce when want the data to stay within your organization or when a specific set of skills is required.

  • One or more vendors that you select from the AWS Marketplace. Vendors provide expertise in specific areas.

  • The Amazon Mechanical Turk workforce. This is the largest workforce, but it should only be used for public data or data that has been stripped of any personally identifiable information.

You can also use automated data labeling to reduce the number of data objects that need to be labeled by a human. Automated data labeling uses active learning to determine if a data object can be labeled by machine or if it needs to be sent to a human worker. For more information, see Using Automated Data Labeling.

The data objects to be labeled are contained in an Amazon S3 bucket. You create a manifest file that describes the location of each object. For more information, see Using Input and Output Data.

The output can be used as the manifest file for another labeling job or as training data for your machine learning models.

You can use this operation to create a static labeling job or a streaming labeling job. A static labeling job stops if all data objects in the input manifest file identified in ManifestS3Uri have been labeled. A streaming labeling job runs perpetually until it is manually stopped, or remains idle for 10 days. You can send new data objects to an active (InProgress) streaming labeling job in real time. To learn how to create a static labeling job, see Create a Labeling Job (API) in the Amazon SageMaker Developer Guide. To learn how to create a streaming labeling job, see Create a Streaming Labeling Job.

Request Syntax

{ "HumanTaskConfig": { "AnnotationConsolidationConfig": { "AnnotationConsolidationLambdaArn": "string" }, "MaxConcurrentTaskCount": number, "NumberOfHumanWorkersPerDataObject": number, "PreHumanTaskLambdaArn": "string", "PublicWorkforceTaskPrice": { "AmountInUsd": { "Cents": number, "Dollars": number, "TenthFractionsOfACent": number } }, "TaskAvailabilityLifetimeInSeconds": number, "TaskDescription": "string", "TaskKeywords": [ "string" ], "TaskTimeLimitInSeconds": number, "TaskTitle": "string", "UiConfig": { "HumanTaskUiArn": "string", "UiTemplateS3Uri": "string" }, "WorkteamArn": "string" }, "InputConfig": { "DataAttributes": { "ContentClassifiers": [ "string" ] }, "DataSource": { "S3DataSource": { "ManifestS3Uri": "string" }, "SnsDataSource": { "SnsTopicArn": "string" } } }, "LabelAttributeName": "string", "LabelCategoryConfigS3Uri": "string", "LabelingJobAlgorithmsConfig": { "InitialActiveLearningModelArn": "string", "LabelingJobAlgorithmSpecificationArn": "string", "LabelingJobResourceConfig": { "VolumeKmsKeyId": "string" } }, "LabelingJobName": "string", "OutputConfig": { "KmsKeyId": "string", "S3OutputPath": "string", "SnsTopicArn": "string" }, "RoleArn": "string", "StoppingConditions": { "MaxHumanLabeledObjectCount": number, "MaxPercentageOfInputDatasetLabeled": number }, "Tags": [ { "Key": "string", "Value": "string" } ] }

Request Parameters

For information about the parameters that are common to all actions, see Common Parameters.

The request accepts the following data in JSON format.


Configures the labeling task and how it is presented to workers; including, but not limited to price, keywords, and batch size (task count).

Type: HumanTaskConfig object

Required: Yes


Input data for the labeling job, such as the Amazon S3 location of the data objects and the location of the manifest file that describes the data objects.

You must specify at least one of the following: S3DataSource or SnsDataSource.

  • Use SnsDataSource to specify an SNS input topic for a streaming labeling job. If you do not specify and SNS input topic ARN, Ground Truth will create a one-time labeling job that stops after all data objects in the input manifest file have been labeled.

  • Use S3DataSource to specify an input manifest file for both streaming and one-time labeling jobs. Adding an S3DataSource is optional if you use SnsDataSource to create a streaming labeling job.

If you use the Amazon Mechanical Turk workforce, your input data should not include confidential information, personal information or protected health information. Use ContentClassifiers to specify that your data is free of personally identifiable information and adult content.

Type: LabelingJobInputConfig object

Required: Yes


The attribute name to use for the label in the output manifest file. This is the key for the key/value pair formed with the label that a worker assigns to the object. The name can't end with "-metadata". If you are running a semantic segmentation labeling job, the attribute name must end with "-ref". If you are running any other kind of labeling job, the attribute name must not end with "-ref".


If you are creating an adjustment or verification labeling job, you must use a different LabelAttributeName than the one used in the original labeling job. The original labeling job is the Ground Truth labeling job that produced the labels that you want verified or adjusted. To learn more about adjustment and verification labeling jobs, see Verify and Adjust Labels.

Type: String

Length Constraints: Minimum length of 1. Maximum length of 127.

Pattern: ^[a-zA-Z0-9](-*[a-zA-Z0-9]){0,126}

Required: Yes


The S3 URI of the file, referred to as a label category configuration file, that defines the categories used to label the data objects.

For 3D point cloud and video frame task types, you can add label category attributes and frame attributes to your label category configuration file. To learn how, see Create a Labeling Category Configuration File for 3D Point Cloud Labeling Jobs.

For all other built-in task types and custom tasks, your label category configuration file must be a JSON file in the following format. Identify the labels you want to use by replacing label_1, label_2,...,label_n with your label categories.


"document-version": "2018-11-28",

"labels": [{"label": "label_1"},{"label": "label_2"},...{"label": "label_n"}]


Note the following about the label category configuration file:

  • For image classification and text classification (single and multi-label) you must specify at least two label categories. For all other task types, the minimum number of label categories required is one.

  • Each label category must be unique, you cannot specify duplicate label categories.

  • If you create a 3D point cloud or video frame adjustment or verification labeling job, you must include auditLabelAttributeName in the label category configuration. Use this parameter to enter the LabelAttributeName of the labeling job you want to adjust or verify annotations of.

Type: String

Length Constraints: Maximum length of 1024.

Pattern: ^(https|s3)://([^/]+)/?(.*)$

Required: No


Configures the information required to perform automated data labeling.

Type: LabelingJobAlgorithmsConfig object

Required: No


The name of the labeling job. This name is used to identify the job in a list of labeling jobs. Labeling job names must be unique within an AWS account and region. LabelingJobName is not case sensitive. For example, Example-job and example-job are considered the same labeling job name by Ground Truth.

Type: String

Length Constraints: Minimum length of 1. Maximum length of 63.

Pattern: ^[a-zA-Z0-9](-*[a-zA-Z0-9]){0,62}

Required: Yes


The location of the output data and the AWS Key Management Service key ID for the key used to encrypt the output data, if any.

Type: LabelingJobOutputConfig object

Required: Yes


The Amazon Resource Number (ARN) that Amazon SageMaker assumes to perform tasks on your behalf during data labeling. You must grant this role the necessary permissions so that Amazon SageMaker can successfully complete data labeling.

Type: String

Length Constraints: Minimum length of 20. Maximum length of 2048.

Pattern: ^arn:aws[a-z\-]*:iam::\d{12}:role/?[a-zA-Z_0-9+=,.@\-_/]+$

Required: Yes


A set of conditions for stopping the labeling job. If any of the conditions are met, the job is automatically stopped. You can use these conditions to control the cost of data labeling.

Type: LabelingJobStoppingConditions object

Required: No


An array of key/value pairs. For more information, see Using Cost Allocation Tags in the AWS Billing and Cost Management User Guide.

Type: Array of Tag objects

Array Members: Minimum number of 0 items. Maximum number of 50 items.

Required: No

Response Syntax

{ "LabelingJobArn": "string" }

Response Elements

If the action is successful, the service sends back an HTTP 200 response.

The following data is returned in JSON format by the service.


The Amazon Resource Name (ARN) of the labeling job. You use this ARN to identify the labeling job.

Type: String

Length Constraints: Maximum length of 2048.

Pattern: arn:aws[a-z\-]*:sagemaker:[a-z0-9\-]*:[0-9]{12}:labeling-job/.*


For information about the errors that are common to all actions, see Common Errors.


Resource being accessed is in use.

HTTP Status Code: 400


You have exceeded an Amazon SageMaker resource limit. For example, you might have too many training jobs created.

HTTP Status Code: 400

See Also

For more information about using this API in one of the language-specific AWS SDKs, see the following: