Services or capabilities described in Amazon Web Services documentation might vary by Region. To see the differences applicable to the China Regions, see Getting Started with Amazon Web Services in China (PDF).

Image Semantic Segmentation

To identify the contents of an image at the pixel level, use an Amazon SageMaker Ground Truth semantic segmentation labeling task. When assigned a semantic segmentation labeling job, workers classify pixels in the image into a set of predefined labels or classes. Ground Truth supports single and multi-class semantic segmentation labeling jobs.

You create a semantic segmentation labeling job using the Ground Truth section of the Amazon SageMaker console or the CreateLabelingJob operation.

Creating a Semantic Segmentation Labeling Job (Console)

You can follow the instructions Create a Labeling Job (Console) to learn how to create a semantic segmentation labeling job in the SageMaker console. In Step 10, choose Image from the Task category drop down menu, and choose Semantic segmentation as the task type.

Ground Truth provides a worker UI similar to the following for labeling tasks. When you create the labeling job with the console, you specify instructions to help workers complete the job and labels that workers can choose from.

Create a Semantic Segmentation Labeling Job (API)

To create a semantic segmentation labeling job, use the SageMaker API operation CreateLabelingJob. This API defines this operation for all Amazon SDKs. To see a list of language-specific SDKs supported for this operation, review the See Also section of CreateLabelingJob.

Follow the instructions on Create a Labeling Job (API) and do the following while you configure your request:

The following is an example of an Amazon Python SDK (Boto3) request to create a labeling job in the US East (N. Virginia) Region. All parameters in red should be replaced with your specifications and resources.

response = client.create_labeling_job(
    LabelingJobName='example-semantic-segmentation-labeling-job,
    LabelAttributeName='label',
    InputConfig={
        'DataSource': {
            'S3DataSource': {
                'ManifestS3Uri': 's3://bucket/path/manifest-with-input-data.json'
            }
        },
        'DataAttributes': {
            'ContentClassifiers': [
                'FreeOfPersonallyIdentifiableInformation'|'FreeOfAdultContent',
            ]
        }
    },
    OutputConfig={
        'S3OutputPath': 's3://bucket/path/file-to-store-output-data',
        'KmsKeyId': 'string'
    },
    RoleArn='arn:aws:iam::*:role/*,
    LabelCategoryConfigS3Uri='s3://bucket/path/label-categories.json',
    StoppingConditions={
        'MaxHumanLabeledObjectCount': 123,
        'MaxPercentageOfInputDatasetLabeled': 123
    },
    HumanTaskConfig={
        'WorkteamArn': 'arn:aws:sagemaker:region:*:workteam/private-crowd/*',
        'UiConfig': {
            'UiTemplateS3Uri': 's3://bucket/path/worker-task-template.html'
        },
        'PreHumanTaskLambdaArn': 'arn:aws:lambda:us-east-1:432418664414:function:PRE-SemanticSegmentation,
        'TaskKeywords': [
            'Semantic Segmentation',
        ],
        'TaskTitle': 'Semantic segmentation task',
        'TaskDescription': 'For each category provided, segment out each relevant object using the color associated with that category',
        'NumberOfHumanWorkersPerDataObject': 123,
        'TaskTimeLimitInSeconds': 123,
        'TaskAvailabilityLifetimeInSeconds': 123,
        'MaxConcurrentTaskCount': 123,
        'AnnotationConsolidationConfig': {
            'AnnotationConsolidationLambdaArn': 'arn:aws:lambda:us-east-1:432418664414:function:ACS-SemanticSegmentation'
        },
    Tags=[
        {
            'Key': 'string',
            'Value': 'string'
        },
    ]
)

Provide a Template for Semantic Segmentation Labeling Jobs

If you create a labeling job using the API, you must supply a worker task template in UiTemplateS3Uri. Copy and modify the following template. Only modify the short-instructions, full-instructions, and header.

Upload this template to S3, and provide the S3 URI for this file in UiTemplateS3Uri.

Semantic Segmentation Output Data

Once you have created a semantic segmentation labeling job, your output data will be located in the Amazon S3 bucket specified in the S3OutputPath parameter when using the API or in the Output dataset location field of the Job overview section of the console.

To learn more about the output manifest file generated by Ground Truth and the file structure the Ground Truth uses to store your output data, see Output Data.

To see an example of an output manifest file for a semantic segmentation labeling job, see 3D Point Cloud Semantic Segmentation Output.