I am trying to extract text from such images but Google Vision API does not seem to recognise majority of the text, Can someone suggest a better alternative? 
Results from Google OCR
Asked
Active
Viewed 557 times
0
Divye Shah
- 747
- 1
- 11
- 24
-
1Does "lower your extremely optimistic expectations" count? – Jongware Apr 17 '18 at 12:02
-
could you please show your output? I just tested your image on google cloud vision demo page and i got most of the text detected and recognized correctly. – flamelite Apr 17 '18 at 12:42
-
@flamelite I added the results, it's good but not enough for me to create a json version of the menu. Could there be some strategy for improving the results – Divye Shah Apr 17 '18 at 14:08
-
@DivyeShah what kind of data do you refer by 'menu'? – flamelite Apr 17 '18 at 14:21
-
@flamelite I meant the a json version of the menu card shown in the picture above – Divye Shah Apr 17 '18 at 14:25
-
So you can use the json version information of vision api. – flamelite Apr 17 '18 at 14:32
-
But it is not accurate enough for me to use that JSON for rendering a menu from that? – Divye Shah Apr 17 '18 at 14:37
1 Answers
1
Your image is 593 by 697 pixels, 0.41 MP. The recommended size for character recognition is 1024 x 768. In fact, OCR requires more resolution to detect characters. You may check relevant details on the Supported Images documentation page.
George
- 1,488
- 1
- 10
- 13