Automatically run 'Grab Text' on all images in Library to make images searchable

  • 3
  • Idea
  • Updated 8 months ago
Having a library of images is great, but I dont always remember when I took them and especially if a free form capture, I cant know where it came from. If you ran Grab Text on all images and added it as a kind of metadata they could be searchable on content which would be immensely helpful and almost make it a kind of wiki like tool where I store the relevant parts of snippets and ideas I find.
Photo of iaingblack

iaingblack

  • 4 Posts
  • 3 Reply Likes

Posted 8 months ago

  • 3
Photo of Paul

Paul

  • 1144 Posts
  • 904 Reply Likes
There are standard EXIF metadata for Title, Description, Caption and Comments plus the ability for SnagIt to write its own custom tags.  The problem would be, what to put where.  It might just be easier to write it all to the description tag and then providea search capability.  

OR

Write it all as a set of library tags, using the existing tag and search functionality
Photo of Joe Morgan

Joe Morgan

  • 5668 Posts
  • 2938 Reply Likes
Agreed, random sentences or text grabbed from images have no file structure.

Daffodils in Spring
Dog bites man
Doomsday
Dreamy 

Would all appear in the same place alphabetically.  But a categorical nightmare I would think.
 
Regards,Joe
Photo of Paul

Paul

  • 1144 Posts
  • 904 Reply Likes
I agree Joe, but as these are being created automatically there's no effort involved and IF tags can be searched in a way that entering "Daff Sprin" will do a fuzzy search and return "Daffodils in Spring"
Photo of iaingblack

iaingblack

  • 4 Posts
  • 3 Reply Likes
Hi, yes. I did think about closing this one after I found that, should I consider doing that? 

But, yes, i dont want this to catalog images or export etc.. etc... Purely so that snagit can keep an internal database of what text is in each image and use that to allow me to search for images with that text inside it. Perhaps others would find automatic tagging with words etc... useful, but it is not part of my original idea, i simply meant 'metadata' as in a silent type of mechanism to track the words.
Photo of Paul

Paul

  • 1144 Posts
  • 904 Reply Likes
The term metadata is a little ambiguous.  Tags are a class of metadata.  The issue is, do they solely exist in the SnagIt library or are they also written to the files so that they can be searched in external tools. I would do both

If the user never saves to png the Library needs to find things quickly, so that's why it needs to be in tags.  If the image goes to PNG either by Save, or drag from tray the information needs to be written to EXIF,  But it will need to go to a standard field for any other app to find it.  Hence the suggestion to use Description.

BUT, TechSmith's track record on being nice to metadata is not good.  Currently if you edit an image with SnagIt, that contains existing metadata, SnagIt nukes it.  I have raised that as an issue.

Tags exist already in SnagIt and they are searchable in SnagIt so the OP's idea could be easily implemented.  It's a brilliant idea.  Why has noone thought of it before? :)  The hours I have spent looking for a file.............


@Ian.  One practical question.  Take the screenshot above, of your comment.  How do you envisage SnagIt parsing that paragraph?  Make each sentence a separate tag and break when it comes to a comma , hyphen or semi or whole colon.  And question or exclamation mark too.
Photo of Paul

Paul

  • 1144 Posts
  • 904 Reply Likes
Hi Ian

I think we're fully aligned on the WHAT.  It's the HOW that is the issue.  I find tags unhelpful because of the time it takes to add them manually.  So mine are completely empty om tens of thousand of images.  If your grab text idea was used to populate the existing tags, that'd be friggin' fantastic.

I realise you were talking about a more covert approach but it's possible that by using existing functionality, the amount of code to write would drop significantly.

Again.  Brilliant idea.
Photo of iaingblack

iaingblack

  • 4 Posts
  • 3 Reply Likes
Thanks Paul :)

I guess I don't mind about the how. I don't envision a tag per word or sentence as it would essentially ruin the other tags a user has set, so less is better. The whole text would be a single tag (assuming no limit on tag sizes which is perhaps unrealistic). That tag (or multiple tags which make up the complete text) could be regexed/compared against the search the user makes. 

Or, no tags. I see it is a SQL Compact DB behind the scenes which holds the Library, just hide the text in a column there and give the user an option to export the text if required via a menu option.
Photo of Paul

Paul

  • 1144 Posts
  • 904 Reply Likes
Yep, ultimately HOW is down to TechSmith