Library module Can auto-tag add-ins work well with a keyword hierarchy?

[PhilBurton]

Operating System:Win 10 Pro
Exact Lightroom Version (Help menu > System Info): LR 7.2

There are several programs like excire, Excire Search - Find your Lightroom Images fast and intuitive , that automatically add keywords to a photo based on image recognition techniques. it appears that these keywords are all "flat file." Is there a practical way to use one of these programs if I also want to have a keyword hierarchy?

Phil Burton

 

[Johan Elzenga]

MyKeyworder supports your hierarchy.

 

[Ian.B]

Thanks for that one Johan; I had never heard of it

 

[rob211]

You might wanna consider the Any Vision plugin for Lr by John Ellis, Any Vision Lightroom Plugin

It uses Google to get them, which are basically keywords in some broad categories like Labels, faces (like funny, sad), locations, etc. But they are stored under those categories in your keyword list in a separate hierarchy, like Any Vision>labels, Any Vision>Faces, so they are segregated and hence don't get in the way of your custom keywords. Any Vision has it's own metadata panel (like IPTC, exif, Location, etc). And Google puts number next to labels to indicate confidence. And the locations are coordinates, so you can go to Google Maps to that location. Mr Ellis makes fantastic plugins, and this is quite well thought out for a regular Lr user, unlike some of the other (and increasingly numerous) auto taggers out there.

Mykeyworder put (IIRC) the keywords into hierarchies if you already had them (like if it ID'd a chair, and you had chair under furniture already, it would use that hieararchy). But it wasn't smart enough to say ID "seat" and also put that under "furniture," if you already had furniture. I didn't use the batch feature though. Whether it works better than Any Vision probably depends on how you wanna do it; try both. Too bad Adobe doesn't have a way to get keywords out of Sensei.

 

[PhilBurton]

You might wanna consider the Any Vision plugin for Lr by John Ellis, Any Vision Lightroom Plugin

It uses Google to get them, which are basically keywords in some broad categories like Labels, faces (like funny, sad), locations, etc. But they are stored under those categories in your keyword list in a separate hierarchy, like Any Vision>labels, Any Vision>Faces, so they are segregated and hence don't get in the way of your custom keywords. Any Vision has it's own metadata panel (like IPTC, exif, Location, etc). And Google puts number next to labels to indicate confidence. And the locations are coordinates, so you can go to Google Maps to that location. Mr Ellis makes fantastic plugins, and this is quite well thought out for a regular Lr user, unlike some of the other (and increasingly numerous) auto taggers out there.

Mykeyworder put (IIRC) the keywords into hierarchies if you already had them (like if it ID'd a chair, and you had chair under furniture already, it would use that hieararchy). But it wasn't smart enough to say ID "seat" and also put that under "furniture," if you already had furniture. I didn't use the batch feature though. Whether it works better than Any Vision probably depends on how you wanna do it; try both. Too bad Adobe doesn't have a way to get keywords out of Sensei.

Rob and everyone else,

Maybe I wasn't too clear in my initial question. Let's say I have an existing, but it is very incomplete compared with some of the Controlled Vocabulary products available. I don't need such a comprehensive hierarchy because I don't do stock photography. Also, given my overall range of subjects, that hierarchy would probably be very limited.

The conventional way to build up that hierarchy when starting out is to do an initial set of keywords and then evolve it over time. What if those keywords were presented to me by a tool that auto-identifies the subject(s)? That approach might help me to build up the hierarchy much faster. But the challenge is two-fold
If the tool suggests keywords that are already in my hierarchy, is Lightroom smart enough to recognize those words?
For those words that are not already in my hierarchy, how easy is it to incorporate them into the hierarchy?

Over time, I would hope that more and more auto-suggested keywords would already be in my hierarchy.

Is my explanation in this post clear? (Should this post been the one to start this thread?) Does this post make sense?

Phil

 

[rob211]

As I said, Any Vision will generate such keywords. It then automatically places them in a separate hierarchy, thus "Any Vision>label>muskrat" and "Any Vision>Faces>confused." You could already have "muskrat" as a keyword, say in "critters>rodents>muskrat." So now you'd have two such tags (like if you had "Construction equipment>crane" and "bird>crane." You could rearrange as necessary. I like this approach, since it's so easy to get false positives with AI tagging.

Mykeyworder is rather different, in that it will present the options like "muskrat" and "confused," and then you have an interactive panel to use those or not. If you do use them, it will just tag "muskrat" under "critters" if it's already there. But it won't put it under "critters" automatically if "muskrat" already exists; it just matches.

Lr Classic isn't smart enough to do anything except try to find faces. CC will tag only online.

It's really easiest to just look at the web examples; describing it is more confusing. Just try them. Both the different approaches are useful in different ways. And consider that with Mykeyworder you could maybe tag just a couple of images from a shoot; then use Lr tools to copy those since often many images have the same content; no need to run them all.

BTW, another way to proceed is to just go more full-auto. If you store and/or share online, then you can let either Adobe's Sensei or Google do the content searching. For example, I've got tons of landscape photos in Lr CC. If somebody wants photos with "branches" they can enter that and get tons of examples; I don't keyword on that term. Or "happy," that's possible too. You can even do it locally; Plex can do such tagging on a server at home. Or Flickr.

 

[PhilBurton]

As I said, Any Vision will generate such keywords. It then automatically places them in a separate hierarchy, thus "Any Vision>label>muskrat" and "Any Vision>Faces>confused." You could already have "muskrat" as a keyword, say in "critters>rodents>muskrat." So now you'd have two such tags (like if you had "Construction equipment>crane" and "bird>crane." You could rearrange as necessary. I like this approach, since it's so easy to get false positives with AI tagging.

Mykeyworder is rather different, in that it will present the options like "muskrat" and "confused," and then you have an interactive panel to use those or not. If you do use them, it will just tag "muskrat" under "critters" if it's already there. But it won't put it under "critters" automatically if "muskrat" already exists; it just matches.

Lr Classic isn't smart enough to do anything except try to find faces. CC will tag only online.

It's really easiest to just look at the web examples; describing it is more confusing. Just try them. Both the different approaches are useful in different ways. And consider that with Mykeyworder you could maybe tag just a couple of images from a shoot; then use Lr tools to copy those since often many images have the same content; no need to run them all.

BTW, another way to proceed is to just go more full-auto. If you store and/or share online, then you can let either Adobe's Sensei or Google do the content searching. For example, I've got tons of landscape photos in Lr CC. If somebody wants photos with "branches" they can enter that and get tons of examples; I don't keyword on that term. Or "happy," that's possible too. You can even do it locally; Plex can do such tagging on a server at home. Or Flickr.

Rob211,

Thank you for this clear explanation. However, I can't go to full auto-tag because a tag word like "train" doesn't work for me as a rail enthusiast, the way "sheep" would not work for the person whose recent post also concerned keyword structures. However, I would like to leverage any tools that I can.

Phil Burton

 

[rob211]

Yeah, that's why I like John Ellis solution. I keep all my very specific (to me) keywords somewhat segregated, but still get a bunch of "mountain" or "beach" type keywords that might come in handy...or not. Sorta best of both worlds.