Enhance OCR accuracy to distinguish currency symbol and number
Currently, in Document Matching, Form Extraction, etc., a currency symbol such as "￥" is recognized as a number"1" or "7" when extracting the amount of money, especially if it is handwritten.
Ideally, improve the accuracy of OCR to distinguish between currency symbol and number.
18
votes
Ryo
shared this idea
-
Ryo
commented
Similar to this, if "-" follows soon after an amount e.g. 6,000- , the number is recognized as minus number. ("-" after numbers is a business habit to prevent someone from manipulating the number.)
It is ideal to recognize numbers that "-" follows as positive numbers. -
Genki Sunayama
commented
Without this improvement most Japanese documents with handwritten amount are not recognized.