Anyone know what is necessary to have in order to use tesseract without installation on Linux? I'm using TTesseractOCR4.
But in general, tesseract is quite unstable... Is there other, more stable free OCR supported with lazarus/fpc?
My OCR requirements are minimal, just to recognize standard English character fonts and numbers. But to do that correctly, not to mix 5 and 3, letters with numbers etc.
That is probably because NN is trained with numbers and letters together, which is big mistake anyway, then all become quite unstable...