Jump to content

OCR Not Happening


Recommended Posts

I've searched the web and this forum for help on this, but haven't found an answer to my problem. I've shot a couple of photos to see how the OCR feature in Evernote works, but it doesn't seem to detect any text.

 

The first image (synced two weeks ago) is of a note pad with handwritten notes. I understand if Evernote can't detect my handwriting, but it hasn't detected any of the printed text (company name, slogan, phone number, etc.) either.

 

The second image (synced six days ago) is of a book cover. The title, author, etc. is in an easy-to-read, plain, sans-serif typeface (see attached image - reduced in size from original). Nothing seems to be detected on this either. From what I've read, it sounds like I should be able to do a search in Evernote for the book title and Evernote will find that note, but it doesn't. I've also tried to search for text from inside the note itself with no luck.

 

 

 

I realize that it takes time for Evernote's servers to do the OCR, but two weeks seems awfully long.

post-240416-0-89959200-1418072628_thumb.

Link to comment
  • Level 5*

Assuming you saved the images as JPG or PNG Image, max size 3000 x 2400px & 300 DPI,  AFAIK OCR should be happening.  Premium users get priority,  but weeks is way too long a time.  If you are a premium user,  I'd suggest raising a support request (see below).  If not,  I'd suggest taking a few pictures at various resolutions and sizes to see if you can find a workable setting. 

Link to comment
  • Level 5*

Hmmn.  That certainly should be processed normally.  Have you searched for 'dreamweaver' or some other text from that image in your database?  It's always possible that Evernote has removed image search for Free users (if that's the level of account that you have),  but I would have expected to see more complaints here if that were the case.

 

I'd suggest uninstalling and reinstalling the application (backup your database and export any local notebooks or unsynced notes to ENEX files before you do so,  just in case..) then,  is search still doesn't find key words,  exit Evernote completely and rename your database folder so it gets rebuilt from the server.  If you have a lot of notes this might take a while...

Link to comment
  • Level 5*

@Analog:  I've seen this before.  I think the EN mobile apps use the EN Service to do searches on OCR'd images, whereas the desktop apps use a local index to do the search.  It seems sometimes that EN Win and EN Mac OCR text indexes don't get updated properly.

 

I suggest that you Submit a BUG report via an EN Support Ticket. In the Support Form, select "Report a bug, crash, or data lost", and start the Ticket Title with "BUG:  " to make it clear.  Reporting a bug should be available to all users, including Free Account owners.

 

 

 

I just discovered that my note has been OCR'd. I just tried searching for "dreamweaver" in the Evernote app on my iPhone, and it found it. Oddly, it doesn't work in the Evernote program I have on both my home computer and work computer.

Link to comment

I think I have this figured out, now:

 

If I search for a note, Evernote will search the text in an image, but if I search within a note, Evernote doesn't search the text in images. At least that's what it's doing in my case. However, prior to forcing Evernote to rebuild the database, it wasn't finding the text in images in either case.

Link to comment

OCR is a difficult beast to tame on the desktop clients. Here's what I've figured out so far:

 

Images get indexed for OCR lightning fast (especially for premium users)... BUT... if you modify a note in any way, including moving it to a new notebook or tagging it, the attachments therein get bumped out of their indexed status, and have to be re-indexed - and invariably, in all cases I have observed, the attachments within those notes stubbornly refuse to be re-indexed on my Windows computer. The only way I have found to reliably re-index everything is to do a fresh sync from the Evernote cloud. In that sense, my findings are identical to yours.

 

EDIT: I forgot that on Windows desktop, one can Ctrl + Click on the "Help" tab to reveal some hidden diagnostic features. When you click on "Fix current note" the attachments appear as indexed once more. Once upon a time I tried, "Fix all notes". That just froze my account and caused me a huge headache. So hold thumbs while selecting that option...

 

The moment I copy a previously indexed image (not indexed at the time) to a new note, that image is synced and indexed real quick. Text in that image is then searchable almost immediately. But not the image contents of the original note.

 

Now when it comes to searching text within images within a single note, I have a simple workaround for you. Stick with me...The great thing about the desktop client is that it adjusts your search as you type in the search bar - unlike the mobile device clients and the web client itself (where you have to hit Enter, "Search" or "Go" to initiate a search). This speeds things up on the desktop, for example, when you type in a word and all notes disappear from your results... you then back up a bit and can adjust your search immediately in real time.

 

If you are able to single out a note with an indexed image, you will see the individual words therein being highlighted as you type. In instances where I had seen this happening, I somehow got the feeling that we were able to search within notes, which you can't (well - yes and no). No you can't hit Ctrl + F and find text in OCRed images. Yes you can search within notes if you are able to single out a note. 

 

First, let me show you a couple of screenshots... In the search result below, I singled out a particular comic strip with a few key words. As I continued to type in words I saw in the comic strip in an arbitrary order, they were all highlighted one by one. In that sense, you can search within a note. 

 

OCR.PNG?dl=1

 

 

Next I decided to tag that comic strip with it's own unique tag "OCR SEARCH". When I clicked on that tag, I had an isolated note. All matched words pertained to just that one image. The problem, though, as I mentioned before, is that when I tagged this note and my desktop client auto-synced, I was no longer able to search the image. I checked the note's information... and the attachment was no longer indexed. Nothing would re-index it. I copied the note and synced. I was able to search that specific image again in the new note.

 

OCR%202.PNG?dl=1

 

 

Besides being the only note in a notebook, the only other way I could think of to isolate a single note amongst several in any context (without tagging it and consequently causing the attachments to lose their index status), was to isolate it with a search operator in the search bar. Below you see I used the "intitle:" syntax to give me that specific note... then I can type in any words I see, which will begin to highlight in real time. I don't know how useful this kind of scenario is anyways, besides just for testing and for curiosity's sake:

 

OCR%203.PNG?dl=1

Link to comment
  • Level 5*

I would call this a bug, and report it.

 

OCR of images by Evernote should NOT, and MUST NOT, be " a difficult beast to tame ".

 

Evernote is supposed to help me -- NOT add additional work.

If Evernote is going to offer OCR and indexing of images, then it must work well, and without additional monitoring/fixing by the user.  Otherwise, I can't rely on it.

 

OCR is a difficult beast to tame on the desktop clients. Here's what I've figured out so far:

 

Images get indexed for OCR lightning fast (especially for premium users)... BUT... if you modify a note in any way, including moving it to a new notebook or tagging it, the attachments therein get bumped out of their indexed status, and have to be re-indexed - and invariably, in all cases I have observed, the attachments within those notes stubbornly refuse to be re-indexed on my Windows computer. The only way I have found to reliably re-index everything is to do a fresh sync from the Evernote cloud. In that sense, my findings are identical to yours.

 

 

Link to comment
  • 3 months later...

The simplest way I have found to re-index a set of notes on Windows desktop is to tag all of them with a temporary, arbitrary tag, sync... then head along to the web client and delete that tag from the tag list. Effecting any change on the web client will then sync through to the Windows desktop client, and consequently one's set of notes is re-indexed and searchable via OCR. 

Link to comment

Archived

This topic is now archived and is closed to further replies.

×
×
  • Create New...