Annotation Missing !!!!!!!!!

[Bradly-s]

Annotation missing when i use local storage by PostgreSQL for 25000 tasks.
How to solve it?

[heidi-humansignal]

Hello there!
Thank you for reaching out and reporting this issue.
I understand that you're experiencing missing annotations when using local storage with a PostgreSQL database for 25,000 tasks. I'd be glad to help you troubleshoot this problem.
To assist you better, could you please provide some additional information:

1. Label Studio Version:

• Which version of Label Studio are you using? Ensuring you're on the latest version can help, as updates often include important fixes.

2. Error Messages:

• Are there any error messages or warnings in the Label Studio server logs when you submit annotations?
• Please check the logs for any ERROR or WARNING entries that might indicate an issue during annotation submission.

3. Annotation Behavior:

• Do the annotations appear to be saved immediately after submission but then disappear later?
• Are the annotations missing from the UI, the database, or both?

Thanks,

Tyler Conlee
Head of Support
HumanSignal

Comment by Tyler Conlee
Workflow Run

[Bradly-s]

1.For the Label Studio Version, i cloned the develop branch of the repository on January 3rd.
2. About the annotation missing:
I have around 25600 images in my local storage, i use PostgreSQL rather than the default SQLite.
I selected all 25600 data and clicked the Retrieve Predictions button. After the label-studio-ml-backend prediction was completed, there was no response from the label studio and the frontend interface did not display the predicted label information.

In my latest experiment, I set ML_TIMEOUT-PREDICT to 3000 (export ML_TIMEOUT-PREDICT=3000), and after the label prediction is completed in the label-studio-ml-backend, it takes a long time for the prediction information to be displayed in the label studio frontend interface.

1. Excuse me, is there any way to speed up the process of displaying predicted labels on the label interface of the label studio frontend after the completion of the label-studio-ml-backend prediction?
2. I selected all 25600 data and clicked the export button in YOLO format, and it takes a long time to export the predicted label information.
   Is there any way to speed up the process?

I start label-studio using the command: python label_studio/manage.py runserver
I start label-studio-ml-backend using the command:
cd label-studio-ml-backend/label_studio_ml/examples/yolo
python _wsgi.py

[heidi-humansignal]

Hello,

1. Excuse me, is there any way to speed up the process of displaying predicted labels on the label interface of the label studio frontend after the completion of the label-studio-ml-backend prediction?

Are you reterviing all the prediction in one go? Maybe try doing batch predictions and that might be better. Also, are you running in docker compose or is it running on the system itself?

2. I selected all 25600 data and clicked the export button in YOLO format, and it takes a long time to export the predicted label information.

I would suggest creating a snapshot and download that first. That should speed up some.

Thank you,
Abu

Comment by Abubakar Saad
Workflow Run