Skip to main content
C
crsouser
Known Participant
April 24, 2015
Question

Lightroom 6 / CC2015 - Facial Recognition Terribly Slow, not using all of CPU or GPU, still keeps going when paused.

  • April 24, 2015
  • 34 replies
  • 38815 views

This is actually a couple of issue but wondering if others are experiencing it and worked through it.

I am using a Dual Xeon CPU 3.2 ghz Mac Pro, Catalog on SSD, Images files on Mirrored Pair, Dedicated GPU (max I can install in my version of Mac Pro) and hardware acceleration enabled in Lightroom.

I watched the Lightroom 6 Facial Recognition tutorial which leaves out a lot of the bulk editing and says basically let it loose on your whole catalog.. NOT recommended.

I started out with a couple small portrait galleries that identified a couple hundred total people to seed facial recognition so it didn't suggest everyone is the first person I confirmed (which it will do otherwise). I have also optimized my catalog multiple times.

I have encountered the following serious performance issues and bugs with Facial Recognition.:

  • Lightroom Facial Recognition goes to a ridiculous crawl after about 2000 images to be confirmed. (i.e. 2000-2300 in a couple hours, 800-1200 in the next 12 hours)
  • Lightroom becomes largely unresponsive after having a fair number of images to be confirmed, even after pausing Address and Facial Recognition. So even selecting 4 rows of images can take 5 minutes with several long pauses.
  • Once I select and click confirm it takes up to 2 minutes to update the "to be confirmed" list again.
  • When I click on an individual at the top of the page, pause facial recognition and address lookup it still continues to "Look for similar faces" [BUG!!!!!!!!] even though all I want to do is just confirm some individuals more quickly in bulk with the images already identified.. not continue to look for more as a work around for the painfully slow responsiveness of the module.

The odd part is that with all of the performance issues Lightroom will not use more than 20-30% of my two Xeon CPUs, barely touches my GPU (<10% CPU, 30% memory), my and no more than 35% of my memory. Computer Temps are also barely above startup temperatures and 15-25 degrees cooler than when I run other applications which will consume my entire CPU and memory if I let it. I have explored Lightroom's settings but seen nothing further I can configure to speed it all up.  I have also attempted the operation on images on the SSD, my drobo (known to be slow), an independent fast disk I have, and a pair of raided disks and have the same issues.

I will also note that all of my other applications seem to continue to operate just fine.. the slowness seems to be contained to the Lightroom application itself.

Lightroom version: 6.0 [1014445]

Operating system: Mac OS 10 Version: 10.10 [3]

Application architecture: x64

Logical processor count: 8

Processor speed: 3.2 GHz

Built-in memory: 18,432.0 MB

Real memory available to Lightroom: 18,432.0 MB

Real memory used by Lightroom: 5,537.5 MB (30.0%)

Virtual memory used by Lightroom: 32,240.6 MB

Memory cache size: 4,342.0 MB

Maximum thread count used by Camera Raw: 8

Camera Raw SIMD optimization: SSE2

Displays: 1) 2048x1152

Graphics Processor Info:

NVIDIA GeForce GTX 285 OpenGL Engine

Check OpenGL support: Passed

Vendor: NVIDIA Corporation

Version: 3.3 NVIDIA-10.0.31 310.90.10.05b12

Renderer: NVIDIA GeForce GTX 285 OpenGL Engine

LanguageVersion: 3.30

Application folder: /Applications/Adobe Lightroom

Library Path: /Users/DryClean/Documents/Lightroom_Catalog/MyCat_LR6.lrcat

Settings Folder: /Users/DryClean/Library/Application Support/Adobe/Lightroom

Anyone have any suggestions?

    This topic has been closed for replies.

    34 replies

    Show previous replies
    D
    djmcm
    Participant
    May 10, 2015

    Same Problem here!

    ADOBE, WHAT IS THE SOLUTION?

    2
    2birds44
    Participant
    May 11, 2015

    Is it possible to permanently turn off the face recognition feature until Adobe corrects the problem?

      G
      genefama
      Known Participant
      May 7, 2015

      ‌A big problem continues to be that once you wait for it to index all your faces you find it missed over half of them. There are many cases where it missed the subject of the photo but managed to find a tiny waiter off in the shadows. I don't know how this will ever get fixed; it seems it'll require an update that lets you rerun the indexing a few times, maybe with different levels of granularity. I doubt that's coming.


      It stands to reason that a system that hands you thousands of false positives for every face can't recognize if something is or isn't a face in over half the cases. Faces in profile or tilted down, especially with the eyes looking down, are bypassed completely. I have directories with a thousand people shots in them, many with multiple people, and instead of LR6 returning an index of 1.5k or so, it gives me 385. I'm not sure how valuable the search advantages will be in this case; I can see it not returning some favorite shots of people.


      Anyway, for anyone looking to get past the spinning wheels, work on one directory at a time. Then once a directory is done, keep it selected so you have the same thumbnails of identified faces in the confirmed area, and control-select the next directory. You won't have to reseed the confirmed faces area this way and things move much faster when you're not working with the entire library. It also helps to click each person in the confirmed faces and work in that view sometimes. It'll return faces of family members of that person and you can rename those. You can watch it focus its top results as you confirm but, annoyingly, it doesn't narrow the total suggestions but actually expands them the more faces you confirm and the fewer correct positives remain. (This seems opposite to how it should work. It should give you fewer faces the more you confirm as it gets a better idea what the person looks like and has fewer shots left unconfirmed of that person.)  Yet, even with the expanded results, faces will still escape it and pop up in other people's results as false positives.


      Bottom line: It's only a little better than manual tagging, and not as thorough because of the poor hit rate of the initial indexing. But it works better if you stick to isolated directories and occasionally individual people. At least that way you don't have to wait for it to re-sort tens of thousands of results with every click.

      B
      barnao
      Participating Frequently
      May 6, 2015

      Same here 360k catalog, been running for about 70% of the time for 1.5 weeks...139k to go

      SSD Catalog, RAID 0 (3 x 4tb/7200RPM)

      Lightroom version: CC 2015.0.1 [ 1018573 ]

      License: Creative Cloud

      Operating system: Windows 8.1 Business Edition

      Version: 6.3 [9600]

      Application architecture: x64

      System architecture: x64

      Logical processor count: 12

      Processor speed: 3.3 GHz (OC 4.0Ghz)

      Built-in memory: 16279.2 MB

      Real memory available to Lightroom: 16279.2 MB

      Real memory used by Lightroom: 4177.4 MB (25.6%)

      Virtual memory used by Lightroom: 4366.8 MB

      Memory cache size: 3050.9 MB

      Maximum thread count used by Camera Raw: 6

      Camera Raw SIMD optimization: SSE2,AVX,AVX2

      System DPI setting: 120 DPI

      Desktop composition enabled: Yes

      Displays: 1) 2560x1600, 2) 3840x2160

      Input types: Multitouch: No, Integrated touch: No, Integrated pen: No, External touch: No, External pen: No, Keyboard: No

      Graphics Processor Info:

      GeForce GTX 780/PCIe/SSE2

      Check OpenGL support: Passed

      Vendor: NVIDIA Corporation

      Version: 3.3.0 NVIDIA 347.52

      Renderer: GeForce GTX 780/PCIe/SSE2

      LanguageVersion: 3.30 NVIDIA via Cg compiler

      Adapter #1: Vendor : 10de

        Device : 1004

        Subsystem : 36041458

        Revision : a1

        Video Memory : 2962

      Adapter #2: Vendor : 1414

        Device : 8c

        Subsystem : 0

        Revision : 0

        Video Memory : 0

      AudioDeviceIOBlockSize: 1024

      AudioDeviceName: Speakers (Realtek High Definition Audio)

      AudioDeviceNumberOfChannels: 2

      AudioDeviceSampleRate: 44100

      Build: Uninitialized

      Direct2DEnabled: false

      GPUDevice: not available

      OGLEnabled: true

      E
      edgduke
      Participating Frequently
      May 6, 2015

      actually, alksaki, it pauses the name-guessing, but still plows through the faces, from what i've found, and just gives you a laundry list of "?" faces.

      i'm FINALLY at 25% to go on 218K photos -- i feel like i'm reaching nirvana or something.

      this is after 2 weeks of continuous process and fighting with my wife, because she thinks i spend too much time with "faces" on the computer -- which i do.

      but you know, if you don't, then you just have a boatload of "?" because subsequent scans can't be matched.

      pathetic -- me and LR6.

      A
      alksaki2
      Participant
      May 6, 2015

      edgduke, thanks for the reply.  Adding names to faces in the past was something I often forgot to do.  Now, we have a way to find those faces.  Right now they are a boatload of ?'s. LOL.  My wife thinks I spend too much time in front of the computer, too.  She will be pleasantly surprised when I can quickly find images of the kids and grandkids.


      I was looking for a way to improve performance when working on images.  When the Face Detection under the Identity Plate is paused, the CPU core usage drops to a much lower level.  So, I think Face Recognition stops looking at ALL the images that have been imported over time.  The 'pausing" let LR6 work a little faster. 

      In the paused state, the People icon below the grid activates "Detecting Faces in Filmstrip".  The People grid appears and populates with faces from the selected folder.  The CPU usage goes way up.  After editing names and deleting unwanted "face areas", the CPU usage goes way down.

      It seems that the computer must stay running while scanning all the images to the very end.  I was about halfway through 254K images and shutdown.  On the next opening of LR, it started scanning all 254K again. Yikes!

      ricardol49133120
      ricardol49133120
      Participant
      May 4, 2015

      for sure it is slow. even aperture, which had not been updated for years, would do this way faster.

      D
      davidh93831741
      Participating Frequently
      April 30, 2015

      ‌I've been using Face recogition in Google Picassa for years. It's 10x better/faster - and free! I can't believe that Adobe after all this time, still can't figure out how to implement a now mature technology that's been around for years. I've been hanging out to have this feature integrated into LR, but sadly disappointed.

        T
        Tanja2014
        Inspiring
        April 30, 2015

        i have the same issues plus these two:

        BUG: Face Recognition splatters Face Tag rectangles all over my Images

        LIGHTROOM 6 Facial recognition runs amok

        LIGHTROOM: facial recognition data not saved - potential data loss

        and as i wrote in another post.. i agree adobe should get help from google to get face recognition right.

          R
          ReLe7395059
          Participating Frequently
          April 30, 2015

          I have excatly the same experiences here.

          1. Tagging is very slow. I type a name and it takes 5-7 seconds to register.

          2. When under one person and confirmed photos, if I change name of wrongly tagged photo. About 70% of the time it just says "added to xxxnn", but does not change the tag. I have to edit keyword tag manually

          3. When under one person I confirm photos, they are added to confirmed. My view scrolls up a lot and I have to painfully slowly to scroll back where I was before confiming. It would be great if I could just view the similar faces view and work on that without the view switching to confirmed photos all the time.

          4. Scrolling is VERY VERY slow.

            E
            edgduke
            Participating Frequently
            April 30, 2015

            how in the world was this version released?

            it's like a massive oxymoron -- "from the same company that brought you photoshop and lightroom, now a virtually unusable and freshman beta of face recognition" --

            -- baffling.

              G
              genefama
              Known Participant
              April 30, 2015

              ‌It's like something from the 1990s.

                G
                genefama
                Known Participant
                April 29, 2015

                Forget it if someone's oriented in profile or has sunglasses on. It offers you cartoons on magazine covers and billboards, patches of lawn, people of multiple races and genders, and never narrows the range of whacky suggestions no matter how many faces you confirm.

                My wife was in the room when it asked me if an Emperor Palpatine PEZ dispenser was her (and she's an attractive woman!). She got pissed off at ME.

                This feature is poorly conceived and executed and an uncharacteric misstep for an otherwise great program—enough so that it shakes my confidence in its future development.

                  C
                  crsouserAuthor
                  Known Participant
                  April 30, 2015

                  To be fair to Adobe, a good facial recognition program should identify all faces in a photograph even if it is a human like face on a Pez dispenser like you mentioned and ABSOLUTELY also do it if you have billboards, magazines, etc in the background. This is how it picks many faces out of a crowd.

                  Also Facial Recognition is a predictive analytic tool that requires training. If you let it go on your entire catalog without some 'training directories' first it will have a horrible ID rate. You need to train it first.  I have used other image recognition programs and they too need to be trained. Also keep in mind that Facial Recognition is step one in the evolution of this.. coming next are identification of "what is a bridge", "What is an car" , "what is a Bear", etc

                  My primary issue is the speed performance of the engine and it not utilizing the resources made available to it.This thread is not intended to be RANT about how you do not like the results of its matching.. please start another thread for that.

                  I posted this Blog Post on some tips about how to work around some of the performance issues as well as described how I initially trained it and though yes it does sometimes think a bunch of leaves or a car ornament is a face... but how I trained it on my 48,000 image set.. I only got about 100 instances of that if I recall right.. so not too bad considering.

                    G
                    genefama
                    Known Participant
                    April 30, 2015

                    I agree it should identify photographic faces on magazine covers and billboards, which is why I specified "cartoon" illustrations. It seems like it should be able to tell the difference between the primitive drawing of a queen on a playing card and a real person, but Adobe's own training video shows it mistaking a playing card for a person. Maybe I'm asking too much, though. I'd be happy if it compared faces across my entire library instead of just within folders, if the results narrowed the suggestions instead of expanding them, and if the spinning wheels would go away.

                    O
                    olivierg73562608
                    Participant
                    April 29, 2015

                    Yes same here - MacPro Late 2013 3.5GHz Hex Core, 16GB, catalog on SSD...

                    Face recognition is painfully slow, seems to run on a single core, and very slow to just go through select faces to be confirmed, which is peculiar as just selecting isn't supposed to be a CPU intensive operation.

                    Same performance observation when running over the whole catalog of 120k pictures, slows down.

                    Generally speaking I need to quit and restart lightroom to get back decent performance in all face recognition related operations.

                    D
                    davidh93831741
                    Participating Frequently
                    April 28, 2015

                    And ditto here too. Great idea, poor implmentation. In it's current state this feature is totally unusable (other than a handful of small galleries). Trying to work with 1000's or 10's of thousands of images grinds the process to a crawl. Constanst screen reloads, lags, freezes etc. Needs major work. Rather fustrating as it's the main reason I upgraded from LR5 which itself was only purchased a few months back. I'm seriously condsidering a request for refund.

                      E
                      edgduke
                      Participating Frequently
                      April 28, 2015

                      Now I'm experiencing data loss.  The faces I've detected just got kicked out -- like 1,500 of them.

                      I called Adobe, waited on hold for a while, and some guy picked up, acting like I was bothering him (??).  When I started explaining my problem, he HUNG UP -- !!

                      This version is junk -- sorry, it is.

                        Show more replies
                        Terms & ConditionsAccessibility statement
                        Using the community Terms of use Privacy Cookie preferences Do not sell or share my personal information ad choicesAdChoices