I have an image which I found on Google. My intention was to once train GOCR with that sample image and then reproduce the results with the knowledge I acquired. I used
gocr -i /tmp/scanned2.jpg -m 2 -p /home/marc/.db/ -m 256 -m 130
to train it. -m 2 to use personal database, -p to specifify where it is located, -m 256 to ignore the usual ocr database, -m 130 to interactively train. After I finished training (which really took a long time to do) I had a look at the result and it was ok. But when I try to reuse my database it does not seem to work. When I use
gocr -i /tmp/scanned2.jpg -m 2 -p /home/marc/.db/ -m 256
which is actually the same command as above only without interactive training of the database it does not recognize many of the things I trained GOCR with in the last session. Shouldn't it a least produce a result that is the same as the result that I got when I trained it?
My system is Linux Mint 3.16.0-38. I installed GOCR via apt-get and have the latest version.
