--
You received this message because you are subscribed to a topic in the Google Groups "ocropus" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/ocropus/cHHcMDbja6A/unsubscribe.
To unsubscribe from this group and all its topics, send an email to ocropus+unsubscribe@googlegroups.com.
To post to this group, send email to ocr...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/ocropus/d2604d8c-9701-446b-a87b-518450ac68b8%40googlegroups.com.
If anyone out there is an expert in recognition of noise free, skew free, graphics free, clean pure line images of Arabic text (single line in binary and/or grayscale) please write to me privately. I do not need page segmentation or preprocessing. I have built that for non commercial use for Arabs and need to add some smart recognition engines for the famous fonts. I need 2 at least. I live in the Silicon Valley.
assalamo alaykomthanks Mr adnan for your response and i hope you are doing well.Actually i need to apply a treatment on each character separately, is that possible?
2016-10-10 5:17 GMT+00:00 Adnan ul Hasan <adnan....@gmail.com>:
There is no function for character segmentation in new OCRopus, which is a line-recognizer. That means that it does recognition on text-line level without needing character segmentation.
On Thursday, 29 September 2016 08:42:23 UTC-7, khadija EL Gajoui wrote:hello everyone,i would like to know which function is responsible of segmenting line to characters? is it in the lstm.py?
--
You received this message because you are subscribed to a topic in the Google Groups "ocropus" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/ocropus/cHHcMDbja6A/unsubscribe.
To unsubscribe from this group and all its topics, send an email to ocropus+u...@googlegroups.com.
To post to this group, send email to ocr...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/ocropus/d2604d8c-9701-446b-a87b-518450ac68b8%40googlegroups.com.