This page is still under construction!

This is an overview, made mainly for developers. You can find typical example files made by me and sent by others. You can pick a example and try to improve gocr. Note that the other examples should be recognized like before your changes or even better.

Users can get a first impression here of how well gocr works.

TOPICS:

1. excellent examples
2. good examples
3. unsorted examples (from other people)

Only errors, where good readable characters are not recognized, or completely wrong recognized are counted. Errors where a "0" (zero) is printed as "O" (like Omega) is not counted. Sometimes missing accents are also not counted. Take it as a rough overview. It is also counted by hand, so it is not very exact.

excellent examples

pbm=black/white, clean scans

testfile ----size---- num_c num_errors remarks
-------- kB x y wc g027 
font1.pbm.gz 40 1623 953 905 7 build by latex+gs 300dpi, 12pt, mixed fonts
font2.pbm.gz 24 1083 637 905 14 build by latex+gs 200dpi, 12pt, mixed fonts

--------------------------------------------------------------------
old overview:
testfile size num_c quality time num_errors remarks
-------- x y ------- p1 p1 p2 p3
g300a1.pbm 703 580 469 + 2s 4 - 0
g300a2.pbm 724 1252 1021 + 5s 2 0 0
g300b1.pbm 1564 277 55 + 4s 0 1 1
g300b2.pbm 599 1319 860 snowy 9s 76 - 40
g300b3.pbm 592 1324 934 snowy 7s 36 2 15
g300c1.pbm 750 2771 2182 + 15s 35 1 14
liebfrau1 2289 3200 1927 + 19s 13 0 8
meraji1 1912 1355 1246 thinn 15s 65 4 40
paraguay1 2617 1375 3280 frame 78s 1000 1 55
p1=gocr0.2.4a3 on P400
p2=recognita+4.0
p3=gocr0.2.5
most errors: connected chars, like fi,ff, italic font
 

unsorted examples

 ... 
 ...