#231 Better Korean Capture Data

[Kori Yi]

4.6.3

First of all, the design is great and 100% what I needed.
I can't speak on the other languages, but considering korean OCR was built in, I was expecting higher quality for image recognition

PROBLEM

• Korean OCR when combining simple and complex letters, even with standard image/text quality
• Standard text quality: text is a typed test, with no visible skew/rotation and is black text on a white background

EXAMPLES

1) Case with text line capture - the most accurate but still having issues with complex vowels (combinations)
2) Text Capture - against difficulty with even separating even simple characters ( 하) when put next to complex characters (는) but not very consistent.. and there's issues with detecting spacing I think with this.

EXPECTATION

• Text OCR to be to be consistent

[Alexandra Allard]

Up !!
Hope they gonna see this and fix it !