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

Dumping Segmentation Masks 1.1. takes way too long #53

Closed
2 tasks done
arasharchor opened this issue Nov 17, 2020 · 2 comments · Fixed by #101
Closed
2 tasks done

Dumping Segmentation Masks 1.1. takes way too long #53

arasharchor opened this issue Nov 17, 2020 · 2 comments · Fixed by #101

Comments

@arasharchor
Copy link

My actions before raising this issue

Expected Behaviour

I have 12 4K images which I have annotated with polygons. When I try to download the annotation with the segmentation mask, it takes a long time. I am still waiting from more 1 hours. The website console seems normal and no errors are issued.

Current Behaviour

To my understanding, changing the points to a mask shall be very fast and should not take a long time.

Steps to Reproduce (for bugs)

  1. upload 12 4K images.
  2. upload annotations for about 3000 polygons.
  3. then try to dump the annotations with segmentation mask.

Context

The images are aerial images and they contain very small to big polygons.

Your Environment

  • Operating System Linux: Ubuntu 16.04
  • Cvat version v1.2.0-alpha
@nmanovic nmanovic transferred this issue from cvat-ai/cvat Nov 17, 2020
@zhiltsov-max
Copy link
Contributor

Hi, could you check server load during exporting? Is it possible that there is high RAM consumption and swapping happens? You can check it, for instance, with docker exec -it cvat htop.

@arasharchor
Copy link
Author

@zhiltsov-max @nmanovic

I checked and found out dumping annotations for the Segmentation Mask utilizes a lot of RAM while monitoring HTOP. That was the reason that it was taking too long and crashed after that.

For instance, I wanted to dump an annotation for an image with the size of 5616x3744px containing polygon annotations.

About 60GB of RAM was used until the download file got ready. Unfortunately, my laptop crashed due to no having enough RAM, so I resorted a server to be able to dump the annotations.

To my understanding converting annotation points to an image format shall not use a lot of RAM memory. So I am still wondering what the reason might be.

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

Successfully merging a pull request may close this issue.

2 participants