Azure Form Recognizer API extract table data incorrect

Dharmveer Kirodiwal 0

Hi,

I am trying to extract table layout from pdf and image using Azure Form Recognized in python. But extracted data is incorrect.

I tried to extract table from pdf then I did not find some table layout.
So, I tried to convert pdf into image then extracted table layout from that image. In this case I got the table layout but some data is incorrect. (Example: extracted "N" where actual value is "NO", extracted "13" where actual value is "(13)").

I have attached the document.

I hope your kind help.

Rohit Mungi 49,111 Reputation points Microsoft Employee Moderator

2023-05-29T14:26:02.1566667+00:00

@Dharmveer Kirodiwal If the quality of the image is unaffected then changing the format from PDF to image will probably give the same results. Are you trying out the prebuilt layout model for this scenario?

I think the document did not get attached with this thread but in general the prebuilt models are trained to extract the all recognized text and numbers. So, if the extraction missed certain characters, then it is most likely that the OCR before the extraction did not recognize the same.

Your answer