PDF Extractor

Question

PDF Extractor

Shambhu Rai 1,411

Hi Expert,

I wanted to use Form Recognizer to load the data from PDF files using pipeline is there any perquisites conditions or criteria to use the data fir curation and further transformation purpose

Shambhu Rai 1,411 Reputation points

2022-04-19T16:55:09.23+00:00

suggestion pls
YutongTie-MSFT 54,011 Reputation points Moderator

2022-04-19T21:06:31.63+00:00

Hello @Shambhu Rai

Thanks for reaching out to us here. Are you asking the commercial use permit for the result you got from Azure Form Recognizer Service?

Regards,
Yutong
Shambhu Rai 1,411 Reputation points

2022-04-19T21:19:31.927+00:00

HI Expert,
No. just wanted to check the the prerequisites or once the data loaded in blob or adf...how we can manage the load or parameters
or how it will work after scannnin.. any lefe cycle example or challanges

https://www.invoicesimple.com/wp-content/uploads/2018/05/InvoiceSimple-PDF-Template.pdf
Shambhu Rai 1,411 Reputation points

2022-04-20T02:08:54.317+00:00

How the data gets loaded after extraction in excel
YutongTie-MSFT 54,011 Reputation points Moderator

2022-04-20T02:12:08.457+00:00

@Shambhu Rai

Do you mean you want to use excel file as input? This is currently not supported. The format for input doc is as below:

Supported file formats: JPEG, PNG, BMP, TIFF, and PDF (text-embedded or scanned). Text-embedded PDFs are best to eliminate the possibility of error in character extraction and location.

Regards,
Yutong
Shambhu Rai 1,411 Reputation points

2022-04-20T02:20:30.267+00:00

Hi Expert,

I am talkin about output. How we can load it in blob or adl or excel and after extraction
YutongTie-MSFT 54,011 Reputation points Moderator

2022-04-20T02:37:55.407+00:00

Hello @Shambhu Rai

I hope I get the right point, if you are asking how the result load into blob, it depends on the feature you are using, most of them you can get the JSON result as below and download it directly.

Give feature General Document(https://learn.microsoft.com/en-us/azure/applied-ai-services/form-recognizer/concept-general-document) as an example, after analyzing, you can see the result in the right panel, and you can download it directly. In case you are not familiar with Form Recognizer Studio, I just made a screenshot guidance for you.

Train your custom model is different, the ocr and label JSON file will be in your blob storage directly as we mentioned above. I will highly recommend you take a try for those feature, and let me know which feature you are interested in, I can share more details about it. Below is all the models we have:

More reference you may be interested in:
https://learn.microsoft.com/en-us/azure/applied-ai-services/form-recognizer/concept-model-overview#model-overview
I hope this helps.

Regards,
Yutong
Shambhu Rai 1,411 Reputation points

2022-04-27T15:26:12.72+00:00

Hi Expert

in the blob structure, I am connecting to Azure database and when i mapped the json i am getting status, ppage, bounding text columns, i want table column header and to map with sql server table column

here is my code

{"status":"succeeded","createdDateTime":"2022-04-27T14:15:46Z","lastUpdatedDateTime":"2022-04-27T14:15:48Z","analyzeResult":{"version":"2.1.0","readResults":[{"page":1,"angle":0,"width":11.6806,"height":8.2639,"unit":"inch","lines":[{"boundingBox":[0.1977,0.8568,1.166,0.8619,1.166,1.0647,0.1977,1.0596],"text":"intertek","appearance":{"style":{"name":"other","confidence":0.878}},"words":[{"boundingBox":[0.1977,0.8619,1.1052,0.8619,1.1103,1.0697,0.2028,1.0647],"text":"intertek","confidence":0.986}]},{"boundingBox":[4.19,1.0151,5.5355,1.0151,5.5355,1.0806,4.19,1.0806],"text":"India CRUDE OIL QUALITY REPORT","appearance":{"style":{"name":"other","confidence":1}},"words":[{"boundingBox":[4.19,1.0151,4.4838,1.0151,4.4838,1.0717,4.19,1.0717],"text":"India","confidence":1},{"boundingBox":[4.509,1.0152,4.7505,1.0152,4.7505,1.0717,4.509,1.0717],"text":"CRUDE","confidence":1},{"boundingBox":[4.7771,1.0151,4.8915,1.0151,4.8915,1.0717,4.7771,1.0717],"text":"OIL","confidence":1},{"boundingBox":[4.9153,1.0151,5.2293,1.0151,5.2293,1.0806,4.9153,1.0806],"text":"QUALITY","confidence":1},{"boundingBox":[5.2558,1.0151,5.5355,1.0151,5.5355,1.0717,5.2558,1.0717],"text":"REPORT","confidence":1}]},{"boundingBox":[0.2028,1.085,0.8821,1.085,0.8821,1.161,0.2079,1.161],"text":"Total Quality. Assured.","appearance":{"style":{"name":"other","confidence":0.878}},"words":[{"boundingBox":[0.2079,1.09,0.3447,1.09,0.3498,1.1661,0.2129,1.1661],"text":"Total","confidence":0.994},{"boundingBox":[0.3599,1.09,0.5932,1.09,0.5932,1.1661,0.365,1.1661],"text":"Quality.","confidence":0.991},{"boundingBox":[0.6084,1.09,0.8821,1.09,0.8821,1.161,0.6084,1.1661],"text":"Assured.","confidence":0.94}]},{"boundingBox":[0.4893,1.2427,0.6554,1.2427,0.6554,1.2984,0.4893,1.2984],"text":"Motel:","appearance":{"style":{"name":"other","confidence":1}},"words":[{"boundingBox":[0.4893,1.2427,0.6554,1.2427,0.6554,1.2984,0.4893,1.2984],"text":"Motel:","confidence":1}]},{"boundingBox":[0.9796,1.2418,1.2381,1.2418,1.2381,1.2984,0.9796,1.2984],"text":"Tony","appearance":{"style":{"name":"other","confidence":1}},"words":[{"boundingBox":[0.9796,1.2418,1.2381,1.2418,1.2381,1.2984,0.9796,1.2984],"text":"Tony","confidence":1}]},{"boundingBox":[4.1871,1.2386,5.4172,1.2386,5.4172,1.3128,4.1871,1.3128],"text":"Grade Name: Tony light crude oil","appearance":{"style":{"name":"other","confidence":1}},"words":[{"boundingBox":[4.1871,1.2387,4.399,1.2387,4.399,1.2984,4.1871,1.2984],"text":"Grade","confidence":1},{"boundingBox":[4.4282,1.2426,4.6542,1.2426,4.6542,1.2984,4.4282,1.2984],"text":"Name:","confidence":1},{"boundingBox":[4.6847,1.2427,4.9071,1.2427,4.9071,1.3128,4.6847,1.3128],"text":"Tony","confidence":1},{"boundingBox":[4.9339,1.2386,5.0874,1.2386,5.0874,1.3128,4.9339,1.3128],"text":"light","confidence":1},{"boundingBox":[5.1128,1.2387,5.3097,1.2387,5.3097,1.2984,5.1128,1.2984],"text":"crude","confidence":1},{"boundingBox":[5.336,1.2386,5.4172,1.2386,5.4172,1.2984,5.336,1.2984],"text":"oil","confidence":1}]},{"boundingBox":[8.5307,1.2386,9.1631,1.2386,9.1631,1.3128,8.5307,1.3128],"text":"Month Reporting:","appearance":{"style":{"name":"other","confidence":1}},"words":[{"boundingBox":[8.5307,1.2386,8.7644,1.2386,8.7644,1.2984,8.5307,1.2984],"text":"Month","confidence":1},{"boundingBox":[8.7956,1.24,9.1631,1.24,9.1631,1.3128,8.7956,1.3128],"text":"Reporting:","confidence":1}]},{"boundingBox":[9.5881,1.2418,9.8167,1.2418,9.8167,1.2984,9.5881,1.2984],"text":"Jan-22","appearance":{"style":{"name":"other","confidence":1}},"words":[{"boundingBox":[9.5881,1.2418,9.8167,1.2418,9.8167,1.2984,9.5881,1.2984],"text":"Jan-22","confidence":1}]},{"boundingBox":[0.4393,1.3658,0.7068,1.3658,0.7068,1.4217,0.4393,1.4217]
How can we take columns headers and data and not any other things or may be we remove it use the file properly.just needs header and data

Expected output:
Motel:India Grade Name:India light crude oil India QUALITY

BL Date
Shambhu Rai 1,411 Reputation points

2022-04-27T22:39:35.333+00:00

Suggestion please

Mike 0

Hi Shambhu Rai,

To use Form Recognizer for extracting data from PDF files within a pipeline for curation and transformation purposes, there are several prerequisites and criteria to keep in mind:

PDF Format:

Ensure that the PDF files are in a machine-readable format. If the PDFs contain scanned images, you may need to apply OCR (Optical Character Recognition) to extract text. Form Recognizer works best with PDFs that have selectable text (not just images).

Form Recognizer Version:

  - Check that you're using the latest version of the Form Recognizer API, as there are ongoing improvements and feature releases. The new layout model (v3.0) offers enhanced capabilities for structured data extraction.
  
  **Supported Languages**:
  
     - Form Recognizer supports multiple languages, but be sure to confirm that the language of the data in the PDF is supported by the version you’re using.
     
     **API Setup**:
     
        - You’ll need an Azure account and access to the Form Recognizer service. Ensure that you have the proper API key and endpoint for the service. You can access these from the Azure portal.

Data Schema:

For effective data curation and transformation, you may need to create or customize a schema for the data you want to extract. The Form Recognizer tool allows you to train models based on your data structure if the out-of-the-box solution doesn’t fully meet your needs.

Pipeline Integration:

Ensure your pipeline (whether it's in Azure Data Factory, Databricks, or another orchestration tool) is set up to interact with the Form Recognizer API. This will typically involve setting up a call to the Form Recognizer API from your pipeline and processing the returned data for further use.

Data Transformation:

  - Once data is extracted, ensure that you’ve set up appropriate transformation steps in your pipeline, such as cleaning the data (removing unnecessary fields, correcting formatting issues) and storing it in a structured format that fits your downstream use cases.

  **File Size Limits**:

     - Keep in mind that Form Recognizer has file size limits for processing PDFs. If your files are large, you may need to split them into smaller chunks or consider using batch processing.

     **Data Privacy and Security**:

        - Ensure you have the necessary compliance and data protection measures in place, especially if the PDFs contain sensitive information. Form Recognizer integrates with Azure’s security features to help protect data, but you should also evaluate your data governance policies.

Let me know if you need additional details on setting this up or if you encounter any issues along the way!

Best regards,
Mike

Answer accepted by question author

0 additional answers

Your answer

Shambhu Rai 1,411 Reputation points

2022-04-19T16:55:09.23+00:00

suggestion pls
YutongTie-MSFT 54,011 Reputation points Moderator

2022-04-19T21:06:31.63+00:00

Hello @Shambhu Rai

Thanks for reaching out to us here. Are you asking the commercial use permit for the result you got from Azure Form Recognizer Service?

Regards,
Yutong
Shambhu Rai 1,411 Reputation points

2022-04-19T21:19:31.927+00:00

HI Expert,
No. just wanted to check the the prerequisites or once the data loaded in blob or adf...how we can manage the load or parameters
or how it will work after scannnin.. any lefe cycle example or challanges

https://www.invoicesimple.com/wp-content/uploads/2018/05/InvoiceSimple-PDF-Template.pdf
Shambhu Rai 1,411 Reputation points

2022-04-20T02:08:54.317+00:00

How the data gets loaded after extraction in excel
YutongTie-MSFT 54,011 Reputation points Moderator

2022-04-20T02:12:08.457+00:00

@Shambhu Rai

Do you mean you want to use excel file as input? This is currently not supported. The format for input doc is as below:

Supported file formats: JPEG, PNG, BMP, TIFF, and PDF (text-embedded or scanned). Text-embedded PDFs are best to eliminate the possibility of error in character extraction and location.

Regards,
Yutong
Shambhu Rai 1,411 Reputation points

2022-04-20T02:20:30.267+00:00

Hi Expert,

I am talkin about output. How we can load it in blob or adl or excel and after extraction
YutongTie-MSFT 54,011 Reputation points Moderator

2022-04-20T02:37:55.407+00:00

Hello @Shambhu Rai

I hope I get the right point, if you are asking how the result load into blob, it depends on the feature you are using, most of them you can get the JSON result as below and download it directly.

Give feature General Document(https://learn.microsoft.com/en-us/azure/applied-ai-services/form-recognizer/concept-general-document) as an example, after analyzing, you can see the result in the right panel, and you can download it directly. In case you are not familiar with Form Recognizer Studio, I just made a screenshot guidance for you.

Train your custom model is different, the ocr and label JSON file will be in your blob storage directly as we mentioned above. I will highly recommend you take a try for those feature, and let me know which feature you are interested in, I can share more details about it. Below is all the models we have:

More reference you may be interested in:
https://learn.microsoft.com/en-us/azure/applied-ai-services/form-recognizer/concept-model-overview#model-overview
I hope this helps.

Regards,
Yutong
Shambhu Rai 1,411 Reputation points

2022-04-27T22:39:35.333+00:00

Suggestion please

Answer 1

Hello @Shambhu Rai

Sure, I can provide you a custom training example in Form Recognizer Studio to see the requirement, life cycle and challenge.

There are some points you should know before:

Language Support, please check if your target language is support here: https://learn.microsoft.com/en-us/azure/applied-ai-services/form-recognizer/language-support
Supported document format, please check if your invoice is good for the format requirement: https://learn.microsoft.com/en-us/azure/applied-ai-services/form-recognizer/concept-model-overview#input-requirements 3. QuickStart guidance, you can refer to this guidance to try our product quick to see if it fulfill your need.
https://learn.microsoft.com/en-us/azure/applied-ai-services/form-recognizer/quickstarts/try-v3-form-recognizer-studio

For the life cycle, the documents you upload will not change or disappear from the blob. After the training, the result will be in JSON file in the same blob as below screenshot.

For challenge, for now I feel Form Recognizer is a good fit for common scenario. Based on my knowledge, some of the customer is suffering from the multipage table is not supported now.

Please check above information and let me know if you have other concern, I am glad to help.

Regards,
Yutong

-Please kindly accept the answer if you feel helpful to help the community, thanks a lot.

Share via

PDF Extractor

0 additional answers

Your answer