URGENT: builtin/latest BREAKS our OCR. We want a further extension on builtin/legacy

[Mishkat B]

We have been using Google Cloud Vision OCR for well over a year. However, the latest update does not detect our document clearly despite being scanned and very clear. It misses certain text/digits from the OCR. The builtin/stable works flawlessly until now. 

We would like either

1. a fix for this issue or

2. an extension on builtin/legacy

Please note, due to the confidentiality of the document, I cannot share the sample here publicly. 

[Mishkat B]

Please not the issue exists with builtin/stable as well

[Tianli Yu]

Hi, Mishkat

Please de-identify the image (by masking the sensitive private information) and send the image to me for debugging.

Thank you! 

[Mishkat B]

Hi Tian 

I have sent you the sample and detected text.

Thanks

[Eileen Wilkerson]

Hello I have experienced the same issue. Please let me know the outcome of this. 

Eileen

[Mishkat B]

sure, I'll let you know as soon as they reply

[Big Data Mining]

Really worried about it!!!! We have the same issue with our application in production!

[s_g]

@bigdatamin, we have tracked down the root cause and are working on an option to revert back to the previous reading order. We will also update the release note in the future when there are significant AI quality upgrades. Thanks!

[Big Data Mining]

Just another point: the legacy model will not be avaiable after August 20 and this huge big problems are not solved! I cannot believe that Google is oferring a product ready for production with this kind of versioning issues, bad communcation with the community and not supporting background compatibility.