How does Box indexing work with OCR'ed documents?
We currently have users OCR their documents before uploading onto Box. Is there any additional setup needed to make sure that the Box search API searches through the text in images inside pdfs? How does the box indexing work with documents that are already OCRed? Do we need to build any custom indexes?
-
Hi Murtza,
I am working with Ktatt on this and what we want basically is to understand how Box.com reads an alreadyOcr'd document and gets the text and how then it stores it. we are trying to build an api that can reference what box.com stores relating to the document and output it to a search response.
Thanks,
Barry
-
Curious if this ever got an answer? I've noted that box doesn't seem to index pdfs with an ocr layer, so the text won't allow the document to show up in a search. I get that Box doesn't OCR images, but since we've already done that leg, we were hopeful it would index the text layer for searching. Interestingly you can use the box preview to highlight and copy text etc just like any other document so it is seeing that there is text.
-
You can store the recognized text as metadata on the file. Our search service will index the file metadata, so you can search for a file based on the recognized text.
-
Hi Murtza, I think we are looking for something a bit quicker and hands off- basically we upload the pdf that has an ocr text layer and like any other document the text is indexed for search. Not looking to use the api, addittional editing steps, etc, just simple upload, box index, and search from the interface. It also looks like from the interface that the metadata option is going to be a difficult sell to enter entire lengthy documents easily. Hopefully this clarifies and maybe you have some other thoughts? Or perhaps this is a future feature coming? Thanks.
Please sign in to leave a comment.
Comments
6 comments