0
$\begingroup$

I used OCR to extract text from an image, but there are some spelling mistakes in it :

The text is as follows :

'gaRBOMATED WATER\n\nSFMEETENED CARBONATED 6\nBSREDERTS: CARBONATED WATER,\nSUGAR. ACIOITY REGULATOR (338),\n\nCFFENE. CONTAINS PERMITTED NATURAL\nCOLOUR (1506) AMD ADDED FLAVOURS QUcTURAL,\nSATIRE: OENTICAL AND ARTIFICIAL PLIVOUREE\n\nCOLA\nl 1187.3 PIRANGUT, TAL. MULSHI,\nGBST. PUME 612111, MAHARASHTRA.\nHELPLINE: 1800- 180-2653\ntet indishetptine@cocs-cola.com\nAUTHORITY OF THE COCA-COLA\n‘COCA-COLA PLAZA, ATLANTA, GA 36313, USA\nme DATE OF MANUFACTURE. BATCH NO. &\nLP CNL. OF ae TAXES}:\nSE BOTTOM OF CAN.\n\nTST Fone Sor MOTHS FROM\nWe, RE WHEN STORED ft.\n\nY PLACE.\nChe coca conn\nnee\n\n| BRA License uo:\n‘ eS wo:\n\n \n\x0c'

I would like to know if there are some NLP models/libraries that I can use to correct spelling mistakes(like correcting gaRBOMATED to CARBONATED

$\endgroup$
3
  • $\begingroup$ Of course there are methods and libraries to identify likely correct spelling, however, their accuracy depends on the possible vocabulary and the amount of errors. If you're dealing with soft drink ingredient lists only, the vocabulary might be sufficiently small but the OCR quality seems to be pretty awful. Don't know whether such a model would correct "BSREDERTS" to "INGREDIENTS", for example. And sorry, I can't suggest a specific library as I don't have experience with that area yet. $\endgroup$ Sep 22, 2020 at 5:30
  • $\begingroup$ Thanks a lot. And yes the OCR quality is not very good. I have to work on that code. $\endgroup$
    – Suraj
    Sep 22, 2020 at 6:42
  • $\begingroup$ Since most blatant errors seem to be at the beginning of lines, it's also possible that the image is partially blurry (if taken from a convex shape, highly possible) and therefore pretty unreadable. In that case, better OCR won't help but better imaging might do. $\endgroup$ Sep 23, 2020 at 8:07

0

You must log in to answer this question.