Hi Jie, It would be possible to show how to ingest a PDF document, not a formatted one, as a raw document to analyze the data. (e.g. clinical trials) and how to store in azure SQL or Synapse using a data factory to feed an ML model to be used in Copilot or any other bot. Thank you.
Found it with chatgpt's help! To get what he had in the beginning, you can run this command from the folder of choice in Git Bash for example: python -m venv pyview
Hi, I am stuck in that bounding box part, how to create the boudind box from the given polygons? can you tell me how can I convert it into x, y, w, h format. Thanks
The polygon coordinates are in Top-Left (1), Top-Right (2), Bottom-Right (3), Bottom-Left (4). The scaling is in inches (if pdf) with the origin (0,0) at the top left of the page. polygon: [ x1, y1, x2,y2, x3,y3, x4,y4 ]
Great video! Thanks for sharing. Can you please share with us your github repo as well? I see that you are importing utility in the invoice extraction code, but I couldn't find it anywhere. Would really appreciate it.
Hey, great video! Which approach do you think is better for extracting specific documents/images patterns (say an ID from a country) in another language (maybe even handwritten) when a user uploads the file and for returing the data to him: Tesseract, Google Cloud Vision OCR, Azure or AWS Textract? How can i make it read the file the user uploaded and return to it the data so he can copy, for example? Many thanks!
Tesseract is mostly used to help simplify formatting output from AWS Textract. If your documents content have the same format, my take would be Azure Document AI > AWS Textract > Google Document AI. Otherwise, any of Vision API would work if you just want to extract text + images.
@@jiejenn Thanks for the response! Could you maybe show us how could we upload something like pdf or jpg of a foreign id (that doesn't have pre-built models) and get back its content?
hi, just a question. I have this project in my bachelor thesis. The pdf files are send to backend(c# .net framework) from frontend(angular) now I that I have list of pdf files in my backend how could I send it to Document Intelligence? I already trained my models and I have blob storage but i just cant figure out and i dont know the next step on how to send it to my custom model?
@@jiejenn thank you for your response, would you happen to have something via code that will do the conversion? Am building a file upload program and want to make it use for .doc files as well. Any guidance is appreciated
What about azure computer vision? I Don´t knnow much about azure, but I thought azure cv was the tool used to extract information from pdf or images. Is this Document AI is some sort of the evolution? Again im new, excuse my ignorance
Some of the features are overlapped, but in summary, Azure vision is used dealing with image process vs Document AI is dealing with documents like forms, receipts, invoices, etc.
You are basically paying for pre trained models and servers to process the requests. Most of the open source libraries don't work well when it comes to extracting fields and tables from forms.
Has anyone been getting the following error? azure.core.exceptions.ResourceNotFoundError: (404) Resource not found Code: 404 Message: Resource not found
I was getting the same error and found the solution. The resource only works when set to these locations: East US, West US2, West Europe, North Central US. So just create a new one and change the endpoint and api_key
