I did a check of using both the new training files and the 3.04 code release and compared accuracy. The results are that the new training files are MUCH more accurate than the old ones. However, if I use the new training files with the 3.03 release, accuracy is still the same, so it appears according to my testing that release 3.04 has not provided any accuracy increase. Are there any other reasons to switch over to the 3.04 release?
On Saturday, July 11, 2015 at 12:14:55 AM UTC-5, Mark Seidner wrote: > > Hi everyone, > I downloaded the latest 3.04 code from git and did a build on Windows, > when I tested on some english files with OEM_TESSERACT_CUBE_COMBINED, there > was no difference in accuracy between Tesseract 3.03 and Tesseract 3.04. I > haven't tried OEM_TESSERACT_ONLY yet, to see if there's any difference > here, still I was surprised that accuracy was the same for both releases. > Also why all the hate for OEM_TESSERACT_CUBE_COMBINED? This combination is > incredibly accurate for camera images. > > Anyone care to comment on why accuracy hasn't improved since the last > release? > -- You received this message because you are subscribed to the Google Groups "tesseract-ocr" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To post to this group, send email to [email protected]. Visit this group at http://groups.google.com/group/tesseract-ocr. To view this discussion on the web visit https://groups.google.com/d/msgid/tesseract-ocr/d37a616a-ed85-4f68-8317-1b7ae25b667a%40googlegroups.com. For more options, visit https://groups.google.com/d/optout.
