What's the purpose of ocrx_cinfo?

kba commented 7 years ago

Spec says

  * ocrx_cinfo should nest inside ocrx_line
  * ocrx_cinfo should contain only x_confs, x_bboxes, and cuts attributes

but not what ocrx_cinfo actually is.

amitdo commented 7 years ago

It's not clear whether Tom really wanted both ocr_cinfo and ocrx_cinfo.

kba commented 7 years ago

IIUC:

ocr_cinfo is a generalized form of word, a placeholder to set character cuts.
ocrx_word is a "word". Quotation marks!, since the engine defines what it means with "word".
ocrx_cinfo would then be something within a line that has distinguishable codepoints that is not a "word" in the sense of the engine.
x_bboxes is only mentioned for ocrx_cinfo
cuts is only mentioned for ocr_cinfo

Since neither ocr_cinfo nor ocrx_cinfo seem to have semantics beyond "can contain character-level coordinates", ocrx_cinfo seems redundant.

amitdo commented 7 years ago

kba / hocr-spec