qurator-spk/dinglehopper

DingleHopper does not create results

stefanCCS opened this issue · 2 comments

stefanCCS commented

Using ocrd, version 2.38.0

I have tried out ocrd-dinglehopper like this:

ocrd-dinglehopper -l DEBUG -I $gtfileGrp,$ocrFileGrp -O $dinglefolder -P textequiv_level line

Somehow, from one of the input files no text is taken:
image

{
    "gt": "OCR-D-SEG-KRAK/OCR-D-SEG-KRAK_4749_007817786_00157.xml",
    "ocr": "OCR-D-TESS-OCR-MOD-04/OCR-D-TESS-OCR-MOD-04-4749_007817786_00157.xml",

    "cer": 4.5,
    "wer": Infinity,

    "n_characters": 56,
    "n_words": 0
}

Any idea?

kba commented

Can you provide the GT and OCR for this file as PAGE-XML? Looks to me like the OCR only recognized . for every character?

stefanCCS commented

Well, this was hint enough - many thanks.
My GT-PAGE does not contain any text (I do not know at this moment ...). Therefore, DingleHopper just shows, what is available.
I will close this issue now.