Closed GoogleCodeExporter closed 9 years ago
In continuation of above, further observations are provided herewith.
Result of tesseractOCR: In the output of bmp file(ybkan-16pt.bmp) used for
training
purpose are furnished below:
Total character sets used: 760 approx.as image
displayed correctly : 255 in which mistakes: 44 nos.
Rubbish .... 505
Note: I had increased no.of classed to 2560 and len 12 in the source code and
using 3-times of same bmp files - rebuilded. No improvement from earlier 512
classes. In other words same output when was used 512 classes and after
increased to
2560 classes.
what is obvious to me is that 255 characters are displayed correctly
(disregarding
44 errors)
As Byte variable holds 0 - 255 so it appears that there are byte values in the
relevant source code of Tesseract that may required to change to integer?. Just
increasing the number of classes did not help
Detailed re-examination of codes are required
With Regards.
Original comment by withbles...@gmail.com
on 9 Oct 2007 at 5:25
This was fixed in an earlier version.
Original comment by theraysm...@gmail.com
on 28 Dec 2008 at 7:42
Original issue reported on code.google.com by
withbles...@gmail.com
on 8 Oct 2007 at 3:23Attachments: