I was testing azure computer vision api latest 3.2 version for my pdf files. for some it working fine but for some pdf files which having itext in it (as see this in pdf's meta info)it is not working fine and returning addition garbage values which is not present on the page, also header of table is keep repeating itself after ending of table row. surprisingly same pdf files was perfectly working fine with previous version of OCR.
I have no much idea about it, my be azure have change some technique or add some extra arguments while requesting and I am not sending with it.
so please suggest solution for this issue if anyone know thanks.
There is some chances from V2 to V3 as this document: https://learn.microsoft.com/en-us/azure/cognitive-services/computer-vision/upgrade-api-versions
Besides above, I don't see any extra value will be added in the new V3.2 API.
If it's not confidential, please share the model version you are using and also a sample to us so that we can reproduce.