Does OCR search rely on using the Evernote Server?

[AndyDent 114]

This is related to issues with searching offline. Different results are being reported and I'm wondering if the OCR searching bypasses local indexing, or varies according to account type.

There are a couple of interesting articles on how indexing works but are many years old

• 2011 blog post on indexing
• 2013 blog post on image recognition

[DTLow 5,749]

Each device maintains it's own search index.

The OCR process runs on the server
It's not a separate "OCR search"; the OCR process makes text available to all devices for inclusion in the search index
An example is an image with text.  The OCR process creates a text file paired with the image.

[PinkElephant 9,015]

With PDFs there is the possibility that a pdf already contains its own OCR data. Up to my knowledge these PDFs will not be OCRed again by the server process, they just take the OCR from the pdf to build the search index. The search index then is device-specific and local.

In my case I am using a Windows and a Mac client as desktop solutions. If the index of say the Windows client got broken, this will not influence search on the Mac, and if I then rebuild the search index on the Win-PC, it will as well only create a new local index used only when searching on the PC. Only when on the web client, search is done on the server data.

[CalS 5,311]

14 hours ago, AndyDent said:

This is related to issues with searching offline. Different results are being reported and I'm wondering if the OCR searching bypasses local indexing, or varies according to account type.

There are a couple of interesting articles on how indexing works but are many years old

• 2011 blog post on indexing
• 2013 blog post on image recognition

No science here, but best I can tell if I OCR something on my PC that is what is used on my PC.  The OCR results aren't sent to the server, the note is OCRd on the server.  The results from the server OCR do not override what I have on the PC. 

If I don't OCR on my PC, the server does it and those results are returned to the PC.  Again, from the other thread, I know this since PDF notes are found on my PC that aren't found on the web, but they should be.

Not sure what happens with downloads to the PC that contain OCRd PDFs.  Assuming the same as for my ScanSnap results.  Sorry for the lack of science but this is what I see in terms of results.