Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Tagging takes forever #20

Closed
derritter88 opened this issue May 18, 2021 · 13 comments
Closed

Tagging takes forever #20

derritter88 opened this issue May 18, 2021 · 13 comments

Comments

@derritter88
Copy link

Hello @marcelklehr ,

I have installed the app last Friday and since then it is stuck with tagging only one use and not all.
Not sure if tagging takes so long or if there is any other issue.

@marcelklehr
Copy link
Member

it is stuck with tagging only one use and not all.

I'm not sure I understand you

@derritter88
Copy link
Author

In my Nextcloud log I see randomly a line saying "regonize for user xxx" but it stuckt with this single user.
No other user is being used.

@marcelklehr
Copy link
Member

Do you see tags being added? If tags are being added, then that's as fast as it can go in a background task. I'm working on a cli command that will do everything at once.

@marcelklehr
Copy link
Member

marcelklehr commented May 18, 2021

The tagging speed depends on your hardware performance specs and the amount of images you have. Raspi and 1TB of images is gonna take a while.

@derritter88
Copy link
Author

Do you see tags being added? If tags are being added, then that's as fast as it can go in a background task. I'm working on a cli command that will do everything at once.

I just had a look at the "real" log -> nextcloud.log and not just the online log display.
I can see that there is a progress - slowly but working.

Regarding hardware specs:
VM
24 GB RAM
8 cores (Epyc 7272)
~ 700 GB cloud storage

So I guess I just need to wait.

@derritter88
Copy link
Author

Just for info: The GUI says `

Warning recognize Classifying photos of user mhepp

`

@marcelklehr
Copy link
Member

v1.3 now comes with a command line interface that allows running the classifier on full speed.

@derritter88
Copy link
Author

Thanks for the information - will do the update right now!

May I ask: Is this application only using the CPU or is it also able to use an existing GPU?

@marcelklehr
Copy link
Member

It currently only uses CPU. But by default the cli will use all your CPUs, you can however also set a custom amount of CPUs.

@marcelklehr
Copy link
Member

marcelklehr commented May 20, 2021

The admin user has access to a tag called "Tagged by recognize" to know which photos have been processed.

@derritter88
Copy link
Author

It currently only uses CPU. But by default the cli will use all your CPUs, you can however also set a custom amount of CPUs.

Would be great if this application would be able to use a GPU.
May I ask which php library your application uses?

Regarding cli: Is it better to create a dedicated cronjob for having the photo recognized or could this be done via the regular Nextcloud cronjob?

@derritter88
Copy link
Author

@marcelklehr regarding my library question:
Unfortunately I am not so deep into development - just remembered some Java stuff I have "developed" at high school.

May I just ask if pdlib would be suitable for this application?
https://github.com/goodspb/pdlib

@marcelklehr
Copy link
Member

Is it better to create a dedicated cronjob for having the photo recognized or could this be done via the regular Nextcloud cronjob?

I think the point of the cli is to do the initial bulk of photo classification, while the background job should be able to handle new photos as they are uploaded. If the background job is not enough, you can schedule the cli to run every night perhaps.

Would be great if this application would be able to use a GPU.

It's definitely possible, you can open a new issue for that.

May I ask which php library your application uses?

I'm using tensorflow-js running inside Node.js.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants