OEM_LSTM_ONLY very slow

[rudrabhoj]

Using Tesseract 4 (libtesseract) in my project decipher_text.
When I use OEM_TESSERACT_ONLY, it takes little over 7-8 seconds to complete the recognization of a page in English. But when I try to instead use OEM_LSTM_ONLY, it takes 10-15 minutes, if not more, to do the same. I am using i5 2400. Ubuntu 16.04 x86_64.

Part of code where this is used:
if (ocrUnit->process->Init(tessDataPath, languageArg, tesseract::OEM_LSTM_ONLY)){
//Again, handle error here
//
//
//
//
}

Entire file in which it was used could be seen here. (Note: File is from an older, stabler version)

[amitdo]

Hi,

Did you test it with the Tesseract command line?
How did you compile it? Did you use debug mode?
Can you provide an image?

[Shreeshrii]

Also see comments in #40 ShreeDevi

…

____________________________________________________________ भजन - कीर्तन - आरती @ http://bhajans.ramparivar.com

On Sun, Jan 1, 2017 at 1:31 PM, Amit D. ***@***.***> wrote: Hi, Did you test it with the Tesseract command line? How did you compile it? Did you use debug mode? Can you provide an image? — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#632 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AE2_o4UvfX-7cir3Ccmjn66P-oI1oZibks5rN11JgaJpZM4LYrS-> .

[theraysmith]

Debug mode is known to be ~50x slower than optimized mode.
Please re-open if you are seeing this slow-down in optimized mode.