Hello all
How well does Ocropus do at recognizing the layout of book pages containing interesting lists and simple tables? Think of a high school yearbook with a creative layout. Perhaps Ocropus won’t identify a column of names that lies diagonal across the page or where the photos and the names make a checker-board pattern. But would it at least identify each name as a separate text line (and let me interpret the bounding box positions)?
If Ocropus has any difficulty with this scenario now (which I’m guessing it will), will it be possible to train Ocropus with hand-labeled pages to learn such page layouts in the near future?
Any other suggestions?
Thanks,
Thomas L. Packer
BYU CS
~~~~~~~~~~~~~~~~~~~~
--
You received this message because you are subscribed to the Google Groups "ocropus" group.
To post to this group, send email to ocr...@googlegroups.com.
To unsubscribe from this group, send email to ocropus+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msg/ocropus/-/r3by0hLJflEJ.
For more options, visit https://groups.google.com/groups/opt_out.