Ready-to-use OCR with 80+ supported languages and all popular writing scripts including Latin, Chinese, Arabic, Devanagari, Cyrillic and etc.
23.82k
stars
3.12k
forks
source link
If my exact string syntax is known, is there a way to pass it as a parameter to improve accuracy? #789
Open
Finatra opened 2 years ago
I need to extract a single numeric string per image.
The strings are formatted.
I know the exact syntax of the strings:
(-)###0.000_(-)###0.000_/_(-)##0.00_(-)##0.00_/_0.000
Where:
(-)
is absent when positive or a minus sign when negative.#
is absent unless it is a number1 through 9
0
is any number0 through 9
(0
is never absent)_
is a space and/
is a forward slash.The string is on a single line and the x/y coordinates of its bounding box are known and constant. The character size and font are known and constant.
Is it possible to inform the model of this syntax? If so, would there be any accuracy or performance advantage?