I'm trying to use the crowd-textract-analyze-document widget on a custom task with sagemaker.
The problem is call the human loop from "start_human_loop" function not from analyze_document.
The human loop tasks is not rendered, the error can be reproduced using "render_ui_template" function
response = sagemaker_client.render_ui_template( UiTemplate={"Content": template_content}, Task={"Input": input_json}, RoleArn=role, )
I'm using the default template from the documentation.
https://docs.aws.amazon.com/sagemaker/latest/dg/a2i-crowd-textract-detection.html
With the minimal ranges to read the data from task.input, so I assume than the error is on the input data.
The error raised is
crowd-html-elements-without-ce-polyfill.js:689 Uncaught TypeError: Cannot read properties of undefined (reading 'text')
The entire json used is this one that includes the "text" attribute, so I don't have any idea of why means the error:
{
"TaskObject": "s3://foo_bar/foo_bar.pdf",
"Keys": [
{
"importantFormKey": "Foo bar"
}
],
"Blocks": [
{
"blockType": "KEY_VALUE_SET",
"confidence": 93.0,
"geometry": {
"boundingBox": {
"width": 0.09730120003223419,
"height": 0.009636270813643932,
"left": 0.5012893676757812,
"top": 0.3701384961605072
},
"polygon": [
{
"x": 0.5012893676757812,
"y": 0.3701384961605072
},
{
"x": 0.5985905528068542,
"y": 0.3701384961605072
},
{
"x": 0.5985905528068542,
"y": 0.3797747790813446
},
{
"x": 0.5012893676757812,
"y": 0.3797747790813446
}
]
},
"id": "6231be31-5f56-41db-95cf-2dad8f765cca",
"relationships": [
{
"type": "VALUE",
"ids": [
"b7697bbd-c5f5-4d28-a345-92c1f53daef7"
]
},
{
"type": "CHILD",
"ids": [
"57c87916-0636-4d51-8d3b-8d15f4e93d73",
"f38b7a0f-2750-46e0-960f-8d6b58dcab3d"
]
}
],
"entityTypes": [
"KEY"
],
"text": "Foo bar",
"page": 1
},
{
"blockType": "KEY_VALUE_SET",
"confidence": 93.0,
"geometry": {
"boundingBox": {
"width": 0.2203546017408371,
"height": 0.019548991695046425,
"left": 0.6025875210762024,
"top": 0.3636907935142517
},
"polygon": [
{
"x": 0.6025875210762024,
"y": 0.3636907935142517
},
{
"x": 0.8229421377182007,
"y": 0.3636907935142517
},
{
"x": 0.8229421377182007,
"y": 0.3832397758960724
},
{
"x": 0.6025875210762024,
"y": 0.3832397758960724
}
]
},
"id": "b7697bbd-c5f5-4d28-a345-92c1f53daef7",
"relationships": [
{
"type": "CHILD",
"ids": [
"d803e2ba-c238-4af1-8745-0dbb17e74420"
]
}
],
"entityTypes": [
"VALUE"
],
"text": "Foo bar",
"page": 1
}
]
}
Hello! Have you checked out this blog post for human-in-the-loop processing of Amazon Textract output? https://aws.amazon.com/blogs/machine-learning/processing-pdf-documents-with-a-human-loop-using-amazon-textract-and-amazon-augmented-ai/
If this doesn't fit your workflow, please let me know and I can dive in more to help address the above error.