Open eroux opened 2 years ago
the following option should be implemented to OCR scans of East-Asian books, that would be in source_metadata
source_metadata
layout
ltr-ttb
ttb-rtl
one example that we will need to OCR is https://library.bdrc.io/show/bdr:W3CN27012
Since we already reorder the polygons, it should be easy to adjust the algorithm with the layout value
layout value
the following option should be implemented to OCR scans of East-Asian books, that would be in
source_metadata
layout
that can be eitherltr-ttb
(left to right, then top to bottom, the usual thing) orttb-rtl
(top to bottom, then left to right)one example that we will need to OCR is https://library.bdrc.io/show/bdr:W3CN27012
Since we already reorder the polygons, it should be easy to adjust the algorithm with the
layout value