aws-samples / aws-ai-intelligent-document-processing

Intelligent Document Processing with AWS AI Services and generative AI

Intelligent Document Processing with AWS AI Services

This repository is part of Intelligent Document Processing with AWS AI Services workshop.

Documents contain valuable information and come in various shapes and forms. In most cases, you are manually processing these documents which is time consuming, prone to error, and expensive. Not only do you want this information extracted quickly but you also want to automate business processes that presently rely on manual inputs and intervention across various file types and formats.

To help you overcome these challenges, AWS Machine Learning (ML) now provides you choices when it comes to extracting information from complex content in any document format such as insurance claims, mortgages, healthcare claims, contracts, and legal contracts.

Different phases of Intelligent Document Processing pipeline

In this workshop, we will deep-dive into each of these phases of the IDP Pipeline with solutions to automate each step. We have hands-on labs to familiarize yourself with AWS AI services ( Amazon Textract, Amazon Comprehend) to build your solution

Getting Started

In order to be able to execute all the Jupyter Notebooks in this sample, we will first need to create a SageMaker Studio domain. The CloudFormation template to create the SageMaker Studio domain and all the related resources, such as IAM Roles, S3 Bucket etc. is included under the /dist directory. Follow the steps below to create the CloudFormation stack using the idp-deploy.yaml file.

⚠️ Your AWS account must have a default VPC for this CloudFormation template to work. Your AWS account may incur some nominal charges for SageMaker Studio domain, Amazon Textract, and Amazon Comprehend. However, Amazon Textract, Comprehend, and SageMaker are free to try as part of AWS Free Tier.

• Navigate to AWS Console
• Search for CloudFormation in the "Services" search bar
• Once in the CloudFormation console, click on the "Create Stack" button (use the "With new resources option")
• In the "Create Stack" wizard, chose "Template is ready", then select "Upload a template file"

• Upload the provided yaml file, click "Next"
• In the "Specify stack details" screen, enter "Stack name". Click "Next"

• In the "Configure Stack options" screen, leave the configurations as-is. Click "Next"
• In the "Review" screen, scroll down to the bottom of the page to the "Capabilities" section and acknowledge the notice that the stack is going to create required IAM Roles by checking the check box. Click "Create stack".

The stack creation can take upto 30 minutes. Once your SageMaker domain is created, you can navigate to the SageMaker console and click on "Amazon SageMaker Studio" on the left pane of the screen. Choose the default user created "SageMakerUser" and Click on "Launch Studio". This will open the SageMaker Studio IDE in a new browser tab. NOTE: If this is your first time using SageMaker Studio then it may take some time for the IDE to fully launch.

Setup SageMaker Studio

Once the SageMaker Studio IDE has fully loaded in your browser, you can clone this repository into the SageMaker Domain instance and start working on the provided Jupyter Notebooks. To clone this repository-

• On the SageMaker Studio IDE, click on "File menu > New > Terminal". This will open a terminal window within SageMaker Studio.

• By default, the terminal launches at the root of the SageMaker Studio IDE workspace.
• Next, clone this repository using

git clone https://github.com/aws-samples/aws-ai-intelligent-document-processing idp_workshop

• Once the repository is cloned, a direcotry named idp_workshop will appear in the "File Browser" on the left panel of SageMaker Studio IDE
• You can now access the Jupyter Notebooks inside the directory and start working on them.

You're all set to begin the workshop!

License

This library is licensed under the MIT-0 License. See the LICENSE file.

aws-ai-intelligent-document-processing's Issues

01-idp-document-classification.ipynb

%store ENDPOINT_ARN
ERROR: UsageError: Unknown variable 'ENDPOINT_ARN'

Cannot Create User Pool for Notebook 04 & 04.01

Launched via Workshop Studio. When creating Private Labelling team:

400 AccessDeniedException: User: arn:aws:sts::xxxxxxxxxx:assumed-role/WSParticipantRole/Participant is not authorized to perform: cognito-idp:CreateUserPool on resource: * because no identity-based policy allows the cognito-idp:CreateUserPool action Request ID: 

CloudFormation error

Despite following the instructions closely, I'm encountering the following error on the first CloudFormation step.
Template error: Unable to get mapping for RegionMap::me-central-1::datascience

AWS::SageMaker::Domain with identifier already exist

HI. During the creation of a new stack I got the message "AWS::SageMaker::Domain with identifier already exist" and the status is automatically set to "rollback". What do I need to change in yaml file to have a new stack created? Thanks

02-idp-document-extraction-01.ipynb

Document

documentName = "simple-document-image.jpg"
display(Image(filename=documentName))

ERROR:

FileNotFoundError Traceback (most recent call last)
in
----> 1 display(Image(filename=documentName))

/opt/conda/lib/python3.7/site-packages/IPython/core/display.py in init(self, data, url, filename, format, embed, width, height, retina, unconfined, metadata)
1230 self.unconfined = unconfined
1231 super(Image, self).init(data=data, url=url, filename=filename,
-> 1232 metadata=metadata)
1233
1234 if self.width is None and self.metadata.get('width', {}):

/opt/conda/lib/python3.7/site-packages/IPython/core/display.py in init(self, data, url, filename, metadata)
635 self.metadata = {}
636
--> 637 self.reload()
638 self._check_data()
639

/opt/conda/lib/python3.7/site-packages/IPython/core/display.py in reload(self)
1261 """Reload the raw data from file or URL."""
1262 if self.embed:
-> 1263 super(Image,self).reload()
1264 if self.retina:
1265 self._retina_shape()

/opt/conda/lib/python3.7/site-packages/IPython/core/display.py in reload(self)
660 """Reload the raw data from file or URL."""
661 if self.filename is not None:
--> 662 with open(self.filename, self._read_flags) as f:
663 self.data = f.read()
664 elif self.url is not None:

FileNotFoundError: [Errno 2] No such file or directory: 'simple-document-image.jpg'

Expected AWS Platform Spend for deployment

Hi,

I'd like to recommend this as a Proof-of-Concept deployment to enable discovery of this service, but to do this I need an estimate of the estimated AWS platform spend for this. Could this be displayed on the README?

Thank you,
@herbtama

02-idp-document-extraction-02.ipynb does not exist

The last step in module2 - doc extraction
cannot performed since the file 02-idp-document-extraction-02.ipynb does not exist.

Fix:
Add and execute the following lines in 02-idp-document-extraction.ipynb to receive the notebook 02-idp-document-extraction-02.ipynb

!wget 'https://github.com/aws-samples/amazon-textract-code-samples/raw/master/python/queries/insurance-card.ipynb' -O './02-idp-document-extraction-02.ipynb'
!wget 'https://github.com/aws-samples/amazon-textract-code-samples/raw/master/python/queries/insurance-card.png' -O './insurance-card.png'