Automate document analysis with Azure Form Recognizer using AI and OCR

แชร์
ฝัง
  • เผยแพร่เมื่อ 2 ส.ค. 2024
  • Extract text automatically from forms, structured or unstructured documents, and text-based images at scale with AI and OCR using Azure’s Form Recognizer service and the Form Recognizer Studio. Build intelligent document processing apps using Azure AI services. Leverage pre-trained models or build your own custom models to help speed up app development with the SDK and APIs.  
    Use a pre-built model for W2 forms
    Build and train custom models to extract data from 1099, 1040, and W9 forms
    Extract data from receipts with handwritten tips, in different languages, currencies, and date formats.
    Bema Bonsu, from Azure's AI engineering team in Azure, joins Jeremy Chapman to share updates to custom app experiences for document processing.
    ► QUICK LINKS:
    00:00 - Introduction
    01:50 - Form Recognizer updates
    03:27 - Extract data from tax forms
    04:56 - Extract data from handwritten text and other languages
    06:00 - Build and train custom models
    08:27 - Extract data from unstructured docs
    10:15 - Code behind custom apps
    12:07 - Wrap up
    ► Link References:
    Get started with Azure Form Recognizer Studio at aka.ms/formrecognizerstudio
    Find out more about Contracts 365 and their services at: www.contracts365.com/
    H&R Block and how they use Form Recognizer: techcommunity.microsoft.com/t...
    ► Unfamiliar with Microsoft Mechanics?
    As Microsoft's official video series for IT, you can watch and share valuable content and demos of current and upcoming tech from the people who build it at Microsoft.
    • Subscribe to our TH-cam: / microsoftmechanicsseries
    • Talk with other IT Pros, join us on the Microsoft Tech Community: techcommunity.microsoft.com/t...
    • Watch or listen from anywhere, subscribe to our podcast: microsoftmechanics.libsyn.com...
    ► Keep getting this insider knowledge, join us on social:
    • Follow us on Twitter: / msftmechanics
    • Share knowledge on LinkedIn: / microsoft-mechanics
    • Enjoy us on Instagram: / msftmechanics
    • Loosen up with us on TikTok: / msftmechanics
    #AzureAI #OpenAI #API #Azure
  • วิทยาศาสตร์และเทคโนโลยี

ความคิดเห็น • 25

  • @perryuploads776
    @perryuploads776 ปีที่แล้ว +6

    This is really wonderfull. This is like OCR next level. Hopefully we get a offline version, that would really be awesome. The AI trained datasets could be downloadable to download specific template for the user. Like how Word or Excel uses for online content and addons. Thanks Microsoft!

  • @abhijithsugunan6768
    @abhijithsugunan6768 ปีที่แล้ว

    Is multipage extraction now available? Last time when I checked if the extracted information lies within two pages, it's not possible with custom model.

  • @agodina
    @agodina ปีที่แล้ว +1

    Fantastic!! Thank you for sharing this.

  • @srikark3532
    @srikark3532 ปีที่แล้ว

    Super!

  • @Abhisingh-cl9xm
    @Abhisingh-cl9xm ปีที่แล้ว

    Is there any code snippet available

  • @imranullah3097
    @imranullah3097 6 หลายเดือนก่อน

    Really great guys. I recommend azure over aws. ❤ Love you 😊

  • @user-tf9pu9ep8h
    @user-tf9pu9ep8h ปีที่แล้ว +3

    I need this entire code to develop the application.Can i get the code? its very useful for my current project

  • @julianlugod
    @julianlugod ปีที่แล้ว

    I hope this will be available offline. I am using Surface Pro 9 SQ3. I hope we can fully utilize with NPU processor.

  • @MountainBiky
    @MountainBiky ปีที่แล้ว

    Hi Microsoft, I have a scenario where we'll be processing invoices and receipts, but we do not know their layout as they will come from a variety of unknown vendors. I'd like to to set the structure myself as to what data to etract such as "vendor name", "vehicle", "License Plate", etc. These invoices may reference these fields with another name, so I don't want form recogniser to start coming up with its own labelling each time. What service would I use in this scenario where I want to spec a structure and we have invoices/recipts coming in from unknown vendors so we can't train on them specifically?

    • @TheBillionDollarSaaS
      @TheBillionDollarSaaS ปีที่แล้ว +1

      Hi, we have a similar usecase. Would you interested in connecting?

    • @danillopinheironeto
      @danillopinheironeto ปีที่แล้ว

      @@TheBillionDollarSaaS @MountainBiky Hello, did you guys got any news about this?
      I have the same usecase.

  • @efhalsey
    @efhalsey ปีที่แล้ว +8

    Is there a GitHub repo for the custom app demo'd?

    • @petersmith8153
      @petersmith8153 ปีที่แล้ว +1

      Yes that would help us recreating it.. Will this be available?

    • @og.m8582
      @og.m8582 ปีที่แล้ว

      Great question

  • @user-jq1fe1bq8t
    @user-jq1fe1bq8t 6 หลายเดือนก่อน

    Bro i got manilla folders with scribble scrabble on them in NO set location. 😂 im so screwed. I don't think this will work

  • @andonii46
    @andonii46 ปีที่แล้ว +1

    How is human validation handled?

    • @MSFTMechanics
      @MSFTMechanics  ปีที่แล้ว +4

      You can build it into your workflow if needed. A good option would be to set thresholds using the confidence percentage.

    • @andonii46
      @andonii46 ปีที่แล้ว

      @@MSFTMechanics shall it be built as a custom web app (for instance)?

    • @notevencare3209
      @notevencare3209 ปีที่แล้ว +1

      So there is pre-built in human-in-the-loop functionality, line Google's Document AI has?

    • @andonii46
      @andonii46 ปีที่แล้ว

      @@notevencare3209 It does not seem so...

    • @ekkamailax
      @ekkamailax 3 หลายเดือนก่อน

      Can the ai hallucinate?

  • @kamaludeenmohamedbatcha3149
    @kamaludeenmohamedbatcha3149 9 หลายเดือนก่อน

    Is this supporting Arabic language

  • @mysterium364
    @mysterium364 10 หลายเดือนก่อน

    6:18

  • @debeerpaul
    @debeerpaul 17 วันที่ผ่านมา

    Please note this can be very very expensive.