Skip navigation

Face recognition

Jan 28, 2009 1:07 AM

  Latest reply: Rob Cole, Mar 31, 2011 7:17 PM
Replies 1 2 Previous Next
  • Currently Being Moderated
    Aug 15, 2010 7:57 PM   in reply to cloveras

    Those who say this is not needed obviously don't have a large collection of family or other people photos to manage.  Let me give the perspective of someone who has a large (more than 31,000) number of photos in Lightroom.  We have been tediously going through adding "tags" or "Keywords" to the metadata for many years.  But in the process, we've obviously missed some in this largely manual process.

     

    I was amazed today to look at my nephew's Apple running iPhoto.  When we added new photos to his collection, it popped up a screen full of faces.  For each, it asked if this was somename or not.  If so, it automatically tagged all obviously related photos with that name keyword.  For mismatches, a click brought up a list of tags or allowed him to add new name keywords.  This is an amazing piece of software that is of great potential value to those of us with serious collections of photos.

     

    Don't dismiss this as trivial just because you might not presently see a use for it.

     
    |
    Mark as:
  • Currently Being Moderated
    Aug 16, 2010 7:07 AM   in reply to RodneyLeeWright

    RodneyLeeWright wrote:

     

    Those who say this is not needed obviously don't have a large collection of family or other people photos to manage.  Let me give the perspective of someone who has a large (more than 31,000) number of photos in Lightroom.

     

    I have over 100,000 in two catalogs.  Face recognition has two major problems.  It doesn't work reliably, at least in Photoshop Elements, and even when it does work it only identifies faces, not people.  I want all the *people* in my images identified, regardless of whether their face appears or not.  So, if I were to keyword my images, I'd have to do it manually anyway.

     
    |
    Mark as:
  • Currently Being Moderated
    Aug 16, 2010 8:36 AM   in reply to Lee Jay-ZyZk56

    I don't really follow that logic, to wit:

    1. No software could identify all of the people in a photo especially if their faces cannot be seen.  So your statement that you'd still have manual processing to do is true.  But evenso, it could still do a LOT of that work with an automated assist...so is still very helpful.  You seem to be saying that since it can't do ALL the work it would not be useful to you.  In my case it would be very useful.

     

    2. I'm not sure what you meant by

    "...it doesn't work properly at least in PhotoShop Elements." I thought we were discussing a potential addin to Adobe Lightroom, not Elements. Personally, I moved to the 64-bit version of Lightroom because Elements was so slow AND extremely unreliable for me.

     

    Sent from my iPhone

     
    |
    Mark as:
  • Currently Being Moderated
    Aug 16, 2010 9:47 AM   in reply to RodneyLeeWright

    RodneyLeeWright wrote:

     

    I don't really follow that logic, to wit:

    1. No software could identify all of the people in a photo especially if their faces cannot be seen.  So your statement that you'd still have manual processing to do is true.  But evenso, it could still do a LOT of that work with an automated assist...so is still very helpful.  You seem to be saying that since it can't do ALL the work it would not be useful to you.  In my case it would be very useful.

     

    2. I'm not sure what you meant by

    "...it doesn't work properly at least in PhotoShop Elements." I thought we were discussing a potential addin to Adobe Lightroom, not Elements. Personally, I moved to the 64-bit version of Lightroom because Elements was so slow AND extremely unreliable for me.

     

    Sent from my iPhone

     

    Correct, but if it didn't do it correctly for all people in all images, then I'd still have to go manually look at each image to see what was done correctly, correct what was done incorrectly, and add what was missed altogether.  I have a hard time believing that would be any faster than just doing the whole thing manually in the first place.

     

    Since Lightroom and Elements are both part of the Photoshop family, I was suggesting that similar base technology might be used for this particular feature.  Since it works so incredibly poorly in Elements 8 (about a 95% failure rate for me), I was implying that it would be worse than worthless to add that to Lightroom.

     

    In short, this technology doesn't seem ready for prime time.  Even if it works as good as in Picasa (which seems to be better), it still fails to identify people who aren't facing the camera, more-or-less directly.

     
    |
    Mark as:
  • Currently Being Moderated
    Aug 16, 2010 12:23 PM   in reply to Lee Jay-ZyZk56

    Hi Rodney,

     

    I think most people would want face recognition IF it worked well and IF it didn't displace too many other features in the coming.

     

    Since Adobe's first attempt doesn't work that well (Photoshop Elements), its clearly a feature that could take a lot more engineering time to get right, so its hard to imagine it working well and not displacing too many other features in the coming.

     

    It strikes me as one of things that should be added only after all the other more important things have been added, like maybe Lr5 or Lr6, but not Lr4. (Unless I over-estimate the amount of effort it would take to get it right).

     

    Rob

     
    |
    Mark as:
  • Currently Being Moderated
    Aug 18, 2010 9:07 AM   in reply to Lee Jay-ZyZk56

    Hi

    I disagree with you. for the following reasons:

    -You are asking the computer to do something that even humans would have difficulties doing: I cannot always tag people on photos whom have hiden faces, especially if I did not take the photo. So I do not really care if the computer does not.

     

    -Face tagging is not here to say who is on the photo but who does a face belong to. When I look at photos 20 years old even if I have the name of all the people on the photo I cannot always say who is who.

     

    -I have sometime been surprised by the accruacy of the software being able to recognise people on photos, among a huge crowd, when myself I did not even notice in many years that I had friends that where there at the same time.

     

    -the time it takes you to tag a face is not proportional to the number of photos you have, but the number of faces you have. So even if you have to check all the photos again, it will go much faster since tagged faces will be highlighted and you won't have to bother for them.

    It takes me at best 5 second to tag a face -> I have to look at the photo find the faces, draw a rectangle around them, and find the name of the person in a huge list of more than a 1000 names or type it. If you say you have 100 000 photos with say 4 faces on average, and a recognition of even a low 5% ( I clearly gets much higher figures around 60%) well it saves you 27 hours of work which makes  3,5 days of work (8h per days). And that is a very pessimistic calculus. Because it very often takes me more than 5 seconds, and the recognition average is higher than 5%and also because even wrong results can be used as explained in the next point:

     

    -Lastly it is not because the software did not properly recognize a face that the results are useless, indeed , if the faces that are found are extracted, groupped by similarities  and displayed in a gallery, like picasa does, then reassigning the correct names is extremely more efficient, than scanning all of your photos. That is the principle strengh of picasa, even when it fails the way it presents the faces makes it extremely easy to assign face tag.

    -figure speak by them selves: it took me a year to do half of my collection (just assigning names not even locating the faces with a rectangle). Using Picasa it took me a week to finish and check that every thing was done.

    My only problem -> it is not compatible with lightroom .

    I am sure lightroom can quickly come to a similar solution. The key point is in organizing the results and the UI for fast tagging and correction.

    Regards

     
    |
    Mark as:
  • Currently Being Moderated
    Aug 18, 2010 11:55 AM   in reply to Babar_e

    Babar_e wrote:

     

    Hi

    I disagree with you. for the following reasons:

    -You are asking the computer to do something that even humans would have difficulties doing: I cannot always tag people on photos whom have hiden faces, especially if I did not take the photo. So I do not really care if the computer does not.

     

     

    I do.  I have many photos like this one that I would want the computer to tag correctly.

     
    |
    Mark as:
  • Currently Being Moderated
    Aug 18, 2010 12:20 PM   in reply to Lee Jay-ZyZk56

    On 8/18/2010 2:55 PM, Lee Jay had this to say:

     
    |
    Mark as:
  • Currently Being Moderated
    Aug 18, 2010 1:51 PM   in reply to Babar_e

    Hi Babar e,

    As you mention that your catalog is tagged in Picasa, there are workarounds to export face data to lightroom. http://regex.info/blog/lightroom-goodies/picasa-face-import has a LR3 plugin that is supposed to do that. I haven't tested it but it might be useful to you. I still hope Adobe introduces this feature in LR4 because without it, i lack the incentive to upgrade to LR3.

     
    |
    Mark as:
  • Currently Being Moderated
    Aug 18, 2010 2:23 PM   in reply to NEF2JPG

    NEF2JPG wrote:

     

    I still hope Adobe introduces this feature in LR4 because without it, i lack the incentive to upgrade to LR3.

     

    So, you're going to wait until Lr4 comes out before deciding to upgrade to Lr3? - Not a bad idea actually, Lr3 will probably be pretty stable by the time Lr4 comes out - ha-ha.

     

    Sorry - I couldn't resist that one...

     

    PS - Despite my understanding of the desire to have the computer tag at least as many faces as it can, alleviating the manual burden, its hard to imagine it being a deal breaker. I mean can you imagine yourself saying:

     

    "I use Brand X - despite its inferior image quality, at least I don't have to manually keyword faces..."?

     

    _R

     
    |
    Mark as:
  • Currently Being Moderated
    Aug 18, 2010 3:29 PM   in reply to Rob Cole

    It's not a deal breaker. It's just that LR3 isn't compelling for me enough as an upgrade over my current setup (LR2 + DxO). For me, LR3's greatest new features are the noise reduction and lens correction. From what i've seen so far, they're not better than DxO and some of my lenses have not been profiled yet. I would highly recommend LR3 standalone but upgrading from LR2 is not so useful if you already have DxO.

     
    |
    Mark as:
  • Currently Being Moderated
    Aug 18, 2010 3:46 PM   in reply to NEF2JPG

    Fair enough. And sorry for being a bit of a smart-a$$.

     

    Personally, I think Lightroom's raw conversion engine and detail handling is better than DxO. Although I like DxO's auto-tone and lens corrections better.

     

    Raw conversion quality + NR + Sharpening + Lens corrections would also top my list of Lr3 advancement over Lr2.

     

    PS - I tried to get DxO to support lens corrections on tif files but to no avail... They claim they need raw info but I don't see why - I mean if the raw file that the tif is based on is still available seems they could just get whatever info they need out of it, skip the raw conversion, and go straight for the lens corrections - but what do I know...

     

    Sorry2 for getting off topic.

     

    PS - I have doubled the ISO I'm willing to shoot at since Lr3 - that's really cool. Although it makes me wonder about myself, since I have Noise Ninja & Topaz DeNoise - I could have just used them with Lr2 and doubled the ISO I was willing to shoot at - but I have a mental block against going outside Lightroom for some reason. Maybe what I need is psycho-therapy instead of better software...

     

    Rob

     
    |
    Mark as:
  • Currently Being Moderated
    Aug 19, 2010 1:55 AM   in reply to Rob Cole

    PS - I have doubled the ISO I'm willing to shoot at since Lr3 - that's really cool. Although it makes me wonder about myself, since I have Noise Ninja & Topaz DeNoise - I could have just used them with Lr2 and doubled the ISO I was willing to shoot at - but I have a mental block against going outside Lightroom for some reason. Maybe what I need is psycho-therapy instead of better software...

     

    I see nothing wrong about trying to maintain a simple and "pure" workflow

     

    I'm considering to buy Topaz DeNoise 5 while the introductory discount lasts, seems to work wonders on JPG. I don't think i'll use it in my own workflow but it would allow me to process some pictures from my wife's compact camera. My psycho-therapy aims to minimize the number of photo software on my laptop and so far it's a big fail

     
    |
    Mark as:
  • Currently Being Moderated
    Aug 19, 2010 2:25 AM   in reply to NEF2JPG

    Sounds like we're both crazy

     

    Tah-tah,

    Rob

     
    |
    Mark as:
  • Currently Being Moderated
    Aug 19, 2010 12:29 PM   in reply to Rob Cole

    Please stay on topic.

     

    Also, please read through threads you reply to. Your concerns have been answered before:

    Face recognition isn't about automatic tagging. Noone would allow a machine to just do that (in Picasa you have to say "yes, that's the person").

     

    What counts is region-tagging, really. The ability to say: HERE, in this photo, you can see X.

     

    If this process is helped along with a face recognition engine - fine, but one keyword with the name is not enough.

     
    |
    Mark as:
  • Currently Being Moderated
    Aug 19, 2010 1:19 PM   in reply to grovel

    grovel wrote:

     

    Please stay on topic.

     

    Sorry about wandering OT

     

    grovel wrote:

     

    Also, please read through threads you reply to. Your concerns have been answered before...

    You talkin' ta me? (I always skim, but sometimes I do cut corners - I admit it)

     

    grovel wrote:

     

    What counts is region-tagging, really. The ability to say: HERE, in this photo, you can see X.

     

    If this process is helped along with a face recognition engine - fine, but one keyword with the name is not enough.

    Thanks for the education - I did not know that.

     

     

    _R

     
    |
    Mark as:
  • Currently Being Moderated
    Aug 24, 2010 2:52 AM   in reply to grovel

    grovel wrote:

     

    Usually, the answer I get when I mention this in forums is: "Lightroom is a tool for professionals" -- I don't think so. It is both priced to appeal to a larger audience, and the people who own a DSLR (most of which are not professionals) do use lightroom or aperture. I realize that 'professionals' usually shoot models (and hence a keyword is enough), but 'amateurs' don't -- for any 'reallife' photography both geo- and face-keywording are essential.

     

    jonathanscottlee wrote:

     

    I second this - definitely need face recognition.  I know a lot of guys are going to be "pros" and will think it's unnecessary but to the advanced amateur, it's mandatory.

    Professionals need three things from their tools, speed, reliability and quality.

    If automatic face tagging can save time then it is a key professional feature. I can think of many jobs where having it would save me a lot of time and effort in keywording, in fact one big long term job, I won't bother with until auto face tagging is possible in LR or to xmp.

    Contrary to popular misbelief, Pros prefer quick and easy to use/understand over complex and hard. Most pro photographers I know regards computers and the attendant software as a necessary evil as they simply make their life more complex, not easier. Particularly when compared to shooting on film. Also Pros have less time to spend learning all the ins and outs of software compared to amateurs, particularly as software changes/updates constantly.

     

    I just downloaded and am trying Picassa's face recognition out. Potentially it is extremely useful from a professional standpoint as well as for personal [family] work, though not with out its failings as it fails on some photos and the options to manually name are missing from interface!?!

    The latest version 3.8 claims to support XMP, but face tagging is different from keywords it would appear. But if you use this Windows only tool AvPicFaceXmpTagger, it can do lots of clever things including it would seem, convert Face tags to keywords. 

     

    Face recognition is a potentially a huge time saver for professional use as well as for any photographer who wants to tag their images.

     
    |
    Mark as:
  • Currently Being Moderated
    Aug 24, 2010 3:03 AM   in reply to imajez

    Whilst testing, Picassa suggested a name for someone I had not yet added to list and the name was the persons mother. Impressive that it can spot family resemblance, but also worrying if the tool will get confused by family members. Hopefully Picassa learns and improves recognition with additional confirmed tags.

     
    |
    Mark as:
  • Currently Being Moderated
    Aug 24, 2010 4:49 AM   in reply to imajez

    Update of face tags to keywords in LR - Cross platform

    A way to keyword people easily from face recognition in Picassa so LR can use info is to select a person in people section, select all the images and add a tag in Picassa, update metadata in LR for individual shots or Sync folders/subfolders and you have keyword for that person added, by virtue of Face recognition.

    Though

     

    Picassa seems to work well with a small no. of people, but the lack of heirachy [as far as I can tell so far] for the people list and inability to group people will be a real pain once you have a lot of people in database. The reason being that being able to select multiple images in untagged faces section and drag them onto a name in people folder is very useful. So if LR is to implement anything like this, being able to place people into folders and several different ones at that, will make it workable. This is very similar to the problem you get with using many Apple designed programmes that are supposedly 'easy' to use. They are only easy if you have little content, once you start to scale up and have a lot of varied content then, they become crippled by their simplistic nature and become very, very difficult to use. Despite using a Mac, I tend to avoid Apple software as it is so slow to use.

     

    A heirachical structure for people, keywords, places etc makes life much easier as being able to select down a list is quicker than finding something in a flat list of say 2,000 items. My big keywording job for a place I've been documenting for several years now will contain a few hundred names for that single documentary alone, but it will be made more manageable by placing names into several categories.  LR's frustrating limitation on only 9 words in a keyword set also makes such a task very challenging as I need keyword sets with 15, 21, 38 etc keywords to be able to keyword efficiently when I am labelling shots with 2-5 people in from a larger subset of say 50 people.

     
    |
    Mark as:
  • Currently Being Moderated
    Feb 24, 2011 3:50 AM   in reply to jonathanscottlee

    I note that after about two years of discussion there is no clue of action on this issue? Have I missed something?

     

    In regard to the comment that Lightroom is for professionals and they don't need this sort of feature I would point out that with the increasing numbers of decent DLSR cameras a number of customers are asking if their 35mm negs can be digitized.

     

    Actually what they don't want is simple digitization - expecially if we're talking a large number of negatives - 'cos that just means they get a load of images on a stick or disc and not a clue as to the data those images contain.Guided updating of the metadata can be all part of the chargeable service.

     
    |
    Mark as:
  • Currently Being Moderated
    Feb 24, 2011 4:25 AM   in reply to drg40

    Please note that I am not a professional, but am in that class labeled prosumers.  That is to say, I am not working for a living as a photographer, but do take many thousands of photos each year...and have converted 10s of thousands of family photos collected over 40 years plus all of the photos of my mother, aunt, and spouse's family.   I also take thousands of aerospace photos for use in a major database at the National Air and Space Museum where I volunteer big time.  My photo database in Lightroom now contains 37,000 photos.

     

    So, does that mean that people like me are not important photographers just because I don't get paid for this?

     

    I have tried using Picassa for facial recognition, then importing the keywords that were created.  It is very, very slow, and hard to use for a Lightroom user.  That said, it worked...almost. As someone else noted here, Picassa's facial recognition does not use a hierarchy.  That means that when I import the keywords from the photo files, they get appended to the end of the keyword list, rather than updating the existing keyword hierarchy (I still call these tags.)  Practically,that means I have had to spend a very large number of hours dragging the imported keywords to their correct location in my hierarchy.  Often I cannot do so as they both have the same name.  Thus an imported keyword for "Jane Austen" has to be renamed to "Jane Austen2" before it can be moved to the correct location.

     

    So, this then reveals the next problem.  I have found no way to merge keywords.  I remember that in Photoshop Elements, there was such a feature.  The lack of that feature in Lightroom 3 means I next have to open all photos of "Jane Austen2" and add the keyword for "Jane Austen" to each record.  Once all "Jane Austen2" keywords are reassigned to "Jane Austen" and the keywords for "Jane Austen2" removed, I delete the now unused "Jane Austen2" keyword.  Make no mistake, this is a procedural kludge.

     

    This whole pain in the bottom could be avoided if Adobe would add facial recogition directly into Lightroom 4 and would add a Merge Keywords function.  I submitted both suggestions to Adobe but of course they don't respond to individual requests other than to acknowledge receipt.  Note that Adobe already knows how to do both of these functions (assuming that Photoshop Lightroom programmers talk to Photoshop Elements programmers.)

     
    |
    Mark as:
  • Currently Being Moderated
    Feb 24, 2011 4:35 AM   in reply to NEF2JPG

    Dumb question time:

    What is DxO?

     
    |
    Mark as:
  • Currently Being Moderated
    Feb 24, 2011 5:07 AM   in reply to RodneyLeeWright

    If I appeared to give the impression that I held amatuer photographers in some sort of contempt I apologise. My point was that there are these days many more people earning their living manipulating photographs than simply those at the low end of Wedding photography. Even for an amateur with the kit a sudden inheritance of 20000 35mm negs comes as something of a blow particularly if you want to save only those photographs of people you recognise. Even worse processing of 120 and all the other film stock sizes which abounded between the wars together with glass plates means that facial recognition is not merely a catch penny as others here have suggested.

     
    |
    Mark as:
  • Currently Being Moderated
    Feb 24, 2011 5:13 AM   in reply to RodneyLeeWright

    I would like to add that face recognition is just an "assist" technology, which is never likely to be 100% accurate and doesn't cover situations where, for example, the person's head is turned away (or even in profile).  However, the key feature that Lightroom misses is a "zone tagging" concept, where just a certain rectangle within the picture can be tagged.  This, combined with some face recognition technology and good tagging management (that is "people aware", would be extremely powerful.  Other pattern-matching technology (to recognise similar forms that are not people faces) would be a very valuable addition to this.

     
    |
    Mark as:
  • Currently Being Moderated
    Feb 24, 2011 5:34 AM   in reply to RodneyLeeWright

    Let's just say DxO is the photo software that offered profile-based lens correction before LR3 started to. And great RAW conversion. And noise reduction too

     

    Those two programs can work together in a photog's workflow but there's so much overlap in functionality that it's probably not worth the hassle anymore if you have LR version 3. I would say the major differences are that DxO does not offer any kind of catalog management (you're just browsing file directories on your hard disk) and is half the price.

     

    To illustrate how i use them, when i was the backup photog for a friend's wedding, i sorted and tweaked each picture individually in LR 2 for optimum results. But when i shoot a typical family outing, i batch-render all pictures in DxO to get better looking JPGs than the camera provided, with zero effort.

     

    As i prefer LR's interface (really a matter of personal opinion), i would probably ditch DxO if i ever upgrade to LR 3. But it's a great, cost-saving alternative if you don't need catalog features, which however doesn't seem to be your case.

     
    |
    Mark as:
  • Currently Being Moderated
    Feb 24, 2011 9:26 AM   in reply to robhylton

    Completely agree with you.

     
    |
    Mark as:
  • Currently Being Moderated
    Mar 1, 2011 1:52 AM   in reply to cloveras

    I second this request.

     
    |
    Mark as:
  • Currently Being Moderated
    Mar 20, 2011 5:48 PM   in reply to cloveras

    I agree with some arguments regarding the release of a stable face tagging solution for Lightroom and I also agree that it is hard to use this as a main criteria for choosing a workflow solution (for advanced and pros). In the other hand, the technology is old but we did not have fast computers before. Even if the detection takes a week, it is one less job we have to do on full manual (besides taking the pictures wich we enjoy, I assume).

     

    As Lightroom is designed for everybody that wants to go a little bit further than pressing a button. I am pretty sure Adobe will loose several customers, me included, not because it lacks the functionallity, but for being unable to keep pace with iPhoto, a software that should be incomparble with Lightroom. What else are we loosing buying LR? That tought is really sad, because I am a big fan of LR wich I have regarded as state-of-tha-art software in its category.

     
    |
    Mark as:
  • Currently Being Moderated
    Mar 31, 2011 7:25 AM   in reply to FaST_5D_Fan

    With all due respect, it seems obvious that there are lots of people who use Lightroom that would like both face recognition and face tagging, of quality at least equal to Picasa.  If the "professionals", whatever that means, don't want to use some tools, they can just ignore them!  I *only* use Lightroom for keywording, ignoring probably 90% of its functionality.

     

    I have taken it on myself to create a database of photos and friends going back to the 19th Century, many digital but many only in hard copy.  One of the main issues is to figure out who the people are, and that really needs to be a collaborative effort among family members all over the world.

     

    Picasa face recognition is far from perfect, but it is on average extremely useful in saving time.  And Picasa face tagging is great.

     

    As has already been noted, when you have thousands of people in a database, a flat list is useless.  I am essentially capturing the geneological tree, plus friends/acquaintances (grouped by any pragmatic method--family member, organizations, places, etc.) in my Lightroom keywords.  I want to be able to tie the faces in the photos to that keyword hierarchy, and make the photo and keyword databases commonly available to stakeholders for revision.  Of course, in addition to the names, I want to share the keyword hierarchies for date, place, subject, event, document type, photographer, etc.

     

    I read that I can save the Picasa face tags in the photos to be read by Lightroom, but the structure will be flat.  Once in Lightroom, I can manually shift the keywords from the flat list to my hierarchy, one name at a time.  But then what happens when I get a new batch of photos? It would seem that I have to do it all again, which would take many days of effort.  This approach is simply unsustainable, and I wish someone would tell me if there is an alternative at any reasonable price.

     

    Also, as I understand it, the Picasa database is configured for use by a single person on a single computer, and the formatting and organization of that database is proprietary and seemingly unavailable.

     

    This is really frustrating, there does not seem to be a tool that is good for both keywording and face recognition/tagging.  I have not looked at iPhoto, and family members run a mixture of Mac OS and Windows, so I prefer not to get something Apple proprietary.  I saw an article online that said that Lightroom keywording is far better than that of iPhoto, but without much specificity about why it is better.  Thoughts?

     

    I just googled "Lightroom 4" and was disappointed and frankly stunned not to see face recognition/tagging in the current list of new features.  Any argument that this functionality is not used by "professionals" AND Lightroom is not intended for anyone who is not a "professional", strikes me as a lame excuse for not investing the required resources.

     

    I strongly encourage Adobe to add facial recognition and facial tagging to Lightroom 4.

     
    |
    Mark as:
  • Currently Being Moderated
    Mar 31, 2011 9:23 AM   in reply to cloveras

    I agree.  I am a professional and I have over 70,000 photos in my Lightroom database.  Some of thos photos are genealogy and a very large number are family and friends.  I would love to have the Facial Recognition feature in Lightroom 4.

     
    |
    Mark as:
  • Currently Being Moderated
    Mar 31, 2011 9:59 AM   in reply to Orrery

    I'm pro, and I need recognition facilities. The reasons for this are simple: to speed-up and enhance catalogization and photography seeking. In my workflow I often assign generic keywords to photos, like «people», «nature» etc. So, if LR could do at least «people» part for me, it would be time-saver. Presently I run Picasa on extracted jpeg previews, and then parse result (just text files) with script. Jeffrey Friedl even wrote plugin, doing the same. But it is not obvious and clear way.

     

    btw, If you're use just 10% of Lightroom, maybe this product is not the best choice for your workflow. But you still have a point.

     
    |
    Mark as:
  • Currently Being Moderated
    Mar 31, 2011 10:39 AM   in reply to alex krylov

    Why? I need the keywording, and use the keyword capability extensively, with many thousands of keywords.  I escape out to Photoshop, which I prefer for editing.  Lightroom's keywording capability for my needs is adequate, not great, but I have found no better alternative yet.

     

    Other than the facial recognition, I only have one other major problem with the keywording--namely, that Lightroom lacks the capability to restrict its display to only the images at a given level in the hierarchy, without showing the images of all subsidiary levels.  So if I ask it to show me all images of my father, it will also display every single image from his side of the family, including all his friends.  Yikes!  There should be a switch to restrict display to a single level. But that is the subject for another thread. 

     
    |
    Mark as:
  • Currently Being Moderated
    Mar 31, 2011 11:02 AM   in reply to Orrery

    It's ok with me, but I think with preambula about 10% LR team can think you're not their target audience and don't schedule this good feature

     
    |
    Mark as:
  • Currently Being Moderated
    Mar 31, 2011 11:52 AM   in reply to alex krylov

        I yam what I yam and that's all what I yam.  (Popeye the Sailor)

     

    Alex,

     

    Your well-taken comment notwithstanding, some retailers prefer to grow their target audiences to increase sales.  And one would think, a heavy user of the product, if not all its functionality, in conjunction with Creative Suite, might not be superfluous.

     

    O.

     
    |
    Mark as:
  • Currently Being Moderated
    Mar 31, 2011 12:42 PM   in reply to Orrery

    Your description of using Picassa for facial recognition and tagging is essentially correct and I have done it.  But because it created a flat file of keywords, it is very cumbersome to import those tags, move them to the correct level in the Lightroom hierarchy, then (because there is no built-in Lightroom facility to merge tags, I have to go through all and manually transfer the tag assignments from one set to another before removing the then empty duplicate tab.  For me it was essentially a one-time process worth doing …but very labor intensive.  A much better solution would be for native Lightroom facial recognition which would eliminate the problem of flat vs. hierarchical tags and the lack of a tag merge function.

     

    Let’s not forget that Adobe already has facial recognition in Elements…therefore that should mean that they already have thoroughly studied how to do this and could apply the logic directly into Elements 4.0.  Of course that assumes that both sets of programmers talk to each other J  Basically I’m saying that inclusion of facial recognition and tagging should be achievable at a reasonable cost for Adobe and they sure would make at least a significant set of their users happy.  Isn’t that supposed to be why they are in business?

     
    |
    Mark as:
  • Currently Being Moderated
    Mar 31, 2011 5:30 PM   in reply to RodneyLeeWright

    Let’s not forget that Adobe already has facial recognition in Elements…therefore that should mean that they already have thoroughly studied how to do this and could apply the logic directly into [Lightroom] 4.0.

    The face recognition in Elements 8 had very serious problems -- see this FAQ:

     

    http://www.johnrellis.com/psedbtool/photoshop-elements-faq.htm#_Face_r ecognition_stops

     

    I haven't tested PSE 9, but I've read reports that it may have similar problems.  The technology in PSE 8 was licensed from a company called Cognitec, and I'd guess that the problem with PSE's recognition was how Adobe applied the technology, rather than the core technology itself.  (But that's an educated guess.)

     
    |
    Mark as:
  • Currently Being Moderated
    Mar 31, 2011 7:17 PM   in reply to Orrery

    Orrery wrote:

     

    ... I only have one other major problem with the keywording--namely, that Lightroom lacks the capability to restrict its display to only the images at a given level in the hierarchy, without showing the images of all subsidiary levels.  So if I ask it to show me all images of my father, it will also display every single image from his side of the family, including all his friends.  Yikes!  There should be a switch to restrict display to a single level. But that is the subject for another thread. 

     

     

    See: http://forums.adobe.com/thread/832328 (the other thread..)

     
    |
    Mark as:
1 2 Previous Next
Actions

More Like This

  • Retrieving data ...

Bookmarked By (0)