[jira] [Resolved] (TIKA-4960) Don't send pages with no text to OCR

["Tim Allison (Jira)" <ji...@apache.org>]

     [ https://issues.apache.org/jira/browse/TIKA-4960?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Tim Allison resolved TIKA-4960.
-------------------------------
    Fix Version/s: 4.2.0
       Resolution: Fixed

> Don't send pages with no text to OCR
> ------------------------------------
>
>                 Key: TIKA-4960
>                 URL: https://issues.apache.org/jira/browse/TIKA-4960
>             Project: Tika
>          Issue Type: Task
>            Reporter: Tim Allison
>            Priority: Minor
>             Fix For: 4.2.0
>
>
> Along with TIKA-4946, it would be great if we could run a lightweight model against an image and decide if it has text before sending it to OCR.
> I'm having mixed success with PP-OCRv6 tiny text detector via oar-ocr's TextDetectionPredictor at this point.
> If anyone wants to jump in with a robust evaluation methodology, please do. I know of several benchmarks, but the models can be trained on those.



--
This message was sent by Atlassian Jira
(v8.20.10#820010)