text detection in vison api is not respecting line segmentation

[surbhi suman]

Hi ,

i m new to to cloud vision api , i am trying to processing receipt having inter word space is greater than normal , and after text detection what i found is vison api is not respecting line segmentation , 

can some one guide how does vision api work  in backend to detect words  in a line .

[Monica (Google Cloud Platform)]

Vision API first performs a layout analysis on the image to segment the location of the text. After the general location is detected, the OCR module then performs a text recognition analysis on the specified location to generate the text. Finally, errors are corrected at a post-processing step by feeding it through a language model or dictionary. You can find more details here.

If you are using document_text_detection you might try text_detection and see if you get better performances. There might be some factors contributing for this issue (like poor image quality or slight angle orientation of the text) and it would be advisable to open a case with support or create a public issue providing your use setup and a sample document (non-confidential).

Since you are handling receipts you might try to test Document AI or Procurement DocAI that are specialized in analyzing invoices.