Oops that one is corrupted, let me try the jpg
Allen Cook
On Wed, Oct 19, 2011 at 3:35 PM, Allen Cook wrote:
> Can anybody give me pointers about how to best improve Tesseract's
> accuracy in this instance? I've attached my current input image to
> Tesseract. This image fails every time, but fre
Can anybody give me pointers about how to best improve Tesseract's
accuracy in this instance? I've attached my current input image to
Tesseract. This image fails every time, but frequently gives me back
different garbage text.
I've also tried training it with a similar font but that never gives
me
Hi,
I was trying to set up the same environment on my Mac. I got some
errors:
Gautam-Mac:tesseract-android-tools Gautam$ ndk-build
Install: libjpeg.so => libs/armeabi/libjpeg.so
Compile++ thumb : lept <= box.cpp
In file included from //Users/Gautam/Documents/workspace/tesseract-
android-
Hey all,
I use Tesseract to automatically OCR batches of TIFF files, but the
accuracy is pretty much hit or miss. I've been using ImageMagick to
convert from PDF to TIFF, and something like "convert -density 380"
will produce great OCR results for one file, whereas the same will not
work well for
Yes, this should certainly be better if you send us one or more example images.
Adding to what have already been said, one thing can be noted for
sure: Tesseract tries to treat everything as a known character, even
schematics or line art. These formations usually appear as garbage in
the output. To
Hi Joao,
You probably need to pre-process the images with other software. By
itself, Tesseract generally cannot ignore data. You can look through
the list archives for some examples. If you provide an example image
(or part of one) it might help us to make a suggestion.
--Sven
On Wed, Oct 19, 201
Normally it's Dmitri's role to say it but I'll do it this time: attach
an image to your question so that we know what you are talking about!
Without it nobody can help.
On Oct 19, 11:42 am, Joao Henriques
wrote:
> Hello everybody,
>
> I hope that someone can help me out here.
> There was nothing
Hello everybody,
I hope that someone can help me out here.
There was nothing on the net regarding it, so'll just try it here :)
I have a picture that needs to be OCR'ed.
The picture contains a schematic and some numbers.
I need to retrieve only these numbers from the picture.
Tesseract keeps try
thanks for reply, then i am going to try something else.
2011/10/18 patrickq
> Unfortunately as far as I know there is no magic setting to tell
> Tesseract to "get spaces right". Ray Smith once wrote that the space
> estimation in Tesseract needs a total rewrite. Within ScanBizCards we
> questi
i do not expect Tesseract for 100% , i want to know what makes the 5%
difference in tesseract can you explain ?
On Tue, Oct 18, 2011 at 7:41 PM, patrickq wrote:
> If you are referring to the confidence level values returned by
> Tesseract these are expressed as "costs" which means a higher values
It look like there are problems with either reading tif image with
tesseract or in image conversion process with ImageMagick.
The image is generated from pdf file and is cropped. I changed from
tif to png format and it looks like the problem is solved.
On Oct 18, 7:45 pm, zdenko podobny wrote:
>
11 matches
Mail list logo