Updated Google Vision OCR API

[Christian Scheier]

Hi,

we notice very different (and for us: pipeline breaking) changes when extracting paragraph bounding boxes in "stable" vs "legacy".

"legacy" works as expected but "stable" returns rather strange bounding boxes. "Stable" appears to maximize (vertical) bounding box size, while "legacy" perfectly matches with expected results (bounding box per paragraph). Example included below.

Word-level bounding boxes still work as expected.

It would be extremely helpful to learn whether this behavior is indeed intended, or what we would do to recover "legacy-like"-behavior for paragraphs.

Thank you

Christian Scheier

[Big Data Mining]

We are facing the same problems!!!! 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.