In reality I have a simple image but the result is always very bad. I tried to binarize but tesseract-OCR should already do it. One result is for example:
'T MRR F 4 4 G 0 4 2 3 0 0 G'.
The image is:
The original is:
Asked
Active
Viewed 30 times
0
-
1Show us the original image. – Nov 05 '19 at 11:36
-
By the way, the results aren't bad at all. The I is dropped and the 1's are consistently confused with 4. The rest is perfect. – Nov 05 '19 at 11:37