Closed sven-nm closed 3 years ago
Looks like a duplicate of https://github.com/OCR-D/ocrd_calamari/issues/57, fixed in v1.0.2 https://github.com/OCR-D/ocrd_calamari/commit/0869386ec4191e7f2b2c96163d5e002eb70c6d35. This is not yet in ocrd_all, though, I'll make a new ocrd_all release this week.
This is fixed in v1.0.2, and the update is also in ocrd_call since last week, as @kba announced here!
@sven-nm I'll close this, please re-open if you still have the issue after updating!
@sven-nm Please note that glyph support in ocrd_calamari is not super useful for image based use cases¹, you might want to explore what ocrd_tesserocr has to offer in this regard.
¹ I implemented it to produce glyph alternatives, but the CTC based coordinates are not suitable for e.g. image extraction
Hi all,
When using
ocrd-calamary-recognize
withtextequiv_level word
,pc:Word
-spans appear to have wrong y-coordinates in theCoords
-spans. It looks like all words are lowered to the bottom of the text region they belong to.For instance : When drawing the line polygons, the coords are right :
But when drawing the word polygons, the coords are wrong :
I am using cv2 to draw the polygons, but I double-checked in the PAGE xml file, and words of a text-region (sometimes the entire page) all have the same y-coordinates.
Here is the entire code used to generate the OCR :