Community

Character extraction in EnginesPool is slow

In the sample EnginesPool, extracting characters after OCR is slow.

Normally, OCR sample RecognizedTextProcessing is fast, so we want to have the same level of speed.

How can I solve it?

 

■Target file

 Demo.tif

 

■Comparison of processing time

 RecognizedTextProcessing → Avg0.18s

 

 EnginesPool → Avg15.2s

 

■Character extraction method

 Implemented in EnginesPool using "calculateStatisticsForLayout" of RecognizedTextProcessing

 

The source ZIP file could not be uploaded...

0

Comments

2 comments

  • Avatar
    OHTSUKA Takeshi

    Someone can explain how to improve the phenomenon that text extraction becomes slow when using EnginesPool because the support of ABBYY Japan isn't useful.
    help me.

    0
  • Avatar
    Ralf Bialozyt

    Hello OHTSUKA,

    Did you get any support meanwhile?

    We are facing a similar issue of slow OCR while using EnginesPool on the Same Intel P7 Hardware.
    For OCR on complex Documents the CPU load is low (even a single Core is not maxed out) while OCR takes sometimes more than 10s.

    0

Please sign in to leave a comment.