Not scanning

[tio-trom]

Which version of recognize are you using?

3.3.3

Enabled Modes

Object recognition, Face recognition, Video recognition

TensorFlow mode

Normal mode

Which Nextcloud version do you have installed?

25.0.2

Which Operating system do you have installed?

Debian 11

Which Docker container are you using to run Nextcloud? (if applicable)

none

How much RAM does your server have?

64GB

What processor Architecture does your CPU have?

AMD64

Describe the Bug

I think I've setup up everything ok but no files are scanned. I use server side encryption and I suspect that's the cause?

I'd like to know. Thanks!

Expected Behavior

My cron is working fine, and I see no logs in nextcloud regarding this issue. I suppose it should start working. I waited for hours now. The CPU is barely used.

To Reproduce

As said above.

Debug log

No response

[ohneb]

Hey @tio-trom, I'm not the developer but I saw in your screenshot that it say in the very top that you did not yet download the models so the recoginition cannot work at this point. You can download them by running occ recognize:download-models on the command line. This will download the models and afterwards it will begin scanning your images.

@marcelklehr Maybe it would be a good idea to the status "The machine learning models still need to be downloaded." as a yellow warning admonition? I was confused by this too when I set up recognize.

Best regards, Benjamin

[bugsyb]

Or potentially to have the models auto-downloaded on the first run? Just a thought, as initially when started with Recognize, that was my impression, gave it a while and then downloaded manually - still unsure if it shouldn't be downloaded automatically.

[Emporea]

I seem to have the same problem. I just setup a new Nextcloud 25 instance yesterday with Preview Generator, Memories and Recognize installed. Somehow I can see some recognized faces but there are 0 tags created so far and nothing has been scanned or classified. I tried downloading the model manually, run scan all from the admin page or recrawl via occ. Nothing seems to work.

[Emporea]

This is part of my config.php. Maybe it has something to do with it?

  'enable_previews' => true,
  'preview_max_x' => 1024,
  'preview_max_y' => 1024,
  'enabledPreviewProviders' => 
  array (
    0 => 'OC\\Preview\\TXT',
    1 => 'OC\\Preview\\MarkDown',
    2 => 'OC\\Preview\\PDF',
    4 => 'OC\\Preview\\JPEG',
    5 => 'OC\\Preview\\PNG',
    6 => 'OC\\Preview\\GIF',
    7 => 'OC\\Preview\\BMP',
    8 => 'OC\\Preview\\XBitmap',
    9 => 'OC\\Preview\\MP3',
    10 => 'OC\\Preview\\HEIC',
    11 => 'OC\\Preview\\Movie',
    12 => 'OC\\Preview\\MKV',
    13 => 'OC\\Preview\\MP4',
    14 => 'OC\\Preview\\AVI',
    18 => 'OC\\Preview\\Krita',
    19 => 'OC\\Preview\\TIFF',
  ),
  'filesystem_check_changes' => 1,

[bugsyb]

What are cron settings? By default it might be that Cron (Ajax) is run on user click/interaction. Give it a try and use php cron.php run out of cron - maybe at the beginning with relatively small interval and see what happens.

Output of oc_jobs table filtered by Recognize could help diagnosing issue too.

[marcelklehr]

Closing due to lack of response.

[tincanfury]

I just ran into the same issue, this post was one of the few hits for Google. Perhaps there needs to just be a note in the installation directions that the occ recognize:download-models needs to be run?

[Emporea]

Recognize is pretty much unusable for me since 3.0.0 cause of several issues.

• classifier trying to allocate more RAM than the php memory_limit of 512mb thus crashing, Fixed this by increaing the memory limit to 1GB.

• 740 ... (This doesnt even get logged for me. I only saw that it failed when running recognize:classify manually in a terminal, after wondering where my clusteres faces should be.

I gave it several chances and let it cluster all of my media. But it failed (me) in the end all the time. Tags are visible, but faces never showed up again since 3.0.0

[marcelklehr]

a note in the installation directions that the occ recognize:download-models needs to be run?

occ recognize:download-models Should not be necessary, recognize should be able to download the models successfully on demand.

classifier trying to allocate more RAM than the php memory_limit

Which process is this? occ recognize:classify? Or a background job?

TypeError: SplObjectStorage::contains():

This will be fixed in the upcoming release, sorry for the inconvenience

But it failed (me) in the end all the time.

Huh, sorry to hear that. Always feel free to open issues with the problems you face.