Problem populating prior labels in adjustment job [AWS Groundtruth labeling]

user_14015778 0
Viewed 214

I'm using the console CLI commands to create new as well adjustment jobs.
I was successfully able to create new jobs for a folder where images are lying but having issues while creating adjustment job to a completed job.

I am using the below command to create the adjustment job

aws sagemaker create-labeling-job --labeling-job-name "test-insulator-relabeling-29-7-2020-v1" --label-attribute-name "insulator-1" --input-config DataSource={S3DataSource={ManifestS3Uri="s3://s3-resized-images/C73/P878813J/OUTPUT/P878813J-objectdetection-2/manifests/output/output.manifest"}},DataAttributes={ContentClassifiers=[""]} --output-config S3OutputPath="s3://prasanna-test-bucket/Sample images folder/OUTPUT/" --role-arn "confidential" --label-category-config-s3-uri "s3://s3-resized-images/Consolidated-Labels/Transformer-Insulator-labels.json" --stopping-conditions MaxPercentageOfInputDatasetLabeled=100 --human-task-config WorkteamArn="confidential",UiConfig={UiTemplateS3Uri="s3://venki-test-raw-images/template1.liquid"},PreHumanTaskLambdaArn="arn:aws:lambda:us-east-1:432418664414:function:PRE-AdjustmentBoundingBox",TaskKeywords="Images","bounding boxes","object detection","adjustment",TaskTitle="test-insulator-relabeling-29-7-2020-v1",TaskDescription="test-insulator-relabeling-29-7-2020-v1",TaskTimeLimitInSeconds=3600,NumberOfHumanWorkersPerDataObject=1,AnnotationConsolidationConfig={AnnotationConsolidationLambdaArn="arn:aws:lambda:us-east-1:432418664414:function:ACS-AdjustmentBoundingBox"}

As soon as I launch the command the job is getting auto completed without the option to adjust the previous jobs labels.
I was able to succesfully created adjustment jobs in AWS UI through browser however the disadvantage there is that i am restricted to previous jobs labels and i cant add any new.

I am trying to figure where I'm going wrong since I made sure all the paramenters required for the adjustment job are given correctly the moment adjustment job launched using CLI as per the above command the job status is getting auto completed.

Can someone help what needs to be understood here to launch to correctly and be able to see previous labels performed in prior job? I am taking care of the config as below

Input : previous jobs output manifest address in S3

label attribute name : same as the prior job

PreHumanTaskLambdaArn="arn:aws:lambda:us-east-1:432418664414:function:PRE-AdjustmentBoundingBox"

AnnotationConsolidationLambdaArn="arn:aws:lambda:us-east-1:432418664414:function:ACS-AdjustmentBoundingBox"

TaskKeywords="Images","bounding boxes","object detection","adjustment"

labeling aws-lambda amazon-sagemaker amazon-web-services amazon-s3
1 Answers

The issue is in the specification of "label attribute name". As the name provided is the same as the previous job, it will be considered as a simple "Chaining" job where we look for remaining objects to be labeled. In your case, as all objects are labeled, the job is auto completing. Please change the label attribute name to a different name and start the above job. When using CLI, you can look at https://docs.aws.amazon.com/sagemaker/latest/dg/sms-verification-data.html and https://docs.aws.amazon.com/sagemaker/latest/dg/sms-ui-template-crowd-bounding-box.html to understand how and where to specify the label attribute name of the prior job. Thank you for being a valued AWS customer.

user_143167190
Related