search

cvat-ai / cvat

Annotate better with CVAT, the industry-leading data engine for machine learning. Used and trusted by teams at any scale, for data of any scale.
https://cvat.ai
MIT License
12.33k stars 2.97k forks source link

does anyone know how to fix loading size #3601

Open SaihholdChiu opened 3 years ago

SaihholdChiu commented 3 years ago

I deployed CVAT on my personal server, i realized it will load all images even I split them to multiple jobs, it caused loading very slowly, so where should I change the loading size to only 1 image?

g-kartik commented 3 years ago

Do you mean, you have created a task with many segment jobs and the uploaded images are large in size.

And when opening a job it downloads all the images of a task even though you have opened only the first job?

SaihholdChiu commented 3 years ago

Do you mean, you have created a task with many segment jobs and the uploaded images are large in size.

And when opening a job it downloads all the images of a task even though you have opened only the first job?

Yes, I created a task with 1 images per job, but when I opened a single job it loaded all images.

g-kartik commented 3 years ago

Yes I confirmed it. When opening any one job, it calls task/id/data with params type=chunk and quality=compressed which returns a zip file with all the images of the task. And it is cached for future requests of job images of the task.

There should be option for the user to set type=frame to get images one by one and not in chunk. And also the chunk should be generated job wise not task wise.

SaihholdChiu commented 3 years ago

Yes I confirmed it. When opening any one job, it calls task/id/data with params type=chunk and quality=compressed which returns a zip file with all the images of the task. And it is cached for future requests of job images of the task.

There should be option for the user to set type=frame to get images one by one and not in chunk. And also the chunk should be generated job wise not task wise.

Yes, at least it should only loads images in the job, instead of the whole task, btw who is this?

azhavoro commented 3 years ago

Hi, it's depends on chunk size option: https://openvinotoolkit.github.io/cvat/docs/manual/basics/creating_an_annotation_task/#chunk-size

SaihholdChiu commented 3 years ago

Hi, it's depends on chunk size option: https://openvinotoolkit.github.io/cvat/docs/manual/basics/creating_an_annotation_task/#chunk-size

So if I set the chunk size as 1, it will load per frame at a single time?

bsekachev commented 3 years ago

So if I set the chunk size as 1, it will load per frame at a single time?

Yes, it is expected.

nmanovic commented 2 years ago

It should load images only for the job. I agree. We will fix that in v2.x with updated REST API.