cvat
cvat copied to clipboard
Support of images with huge width and height
Hello,I submit a bmp file to cvat with the size of 18000*14000 and I got the error
Could not create the task Error: Could not create the task on the server. PIL.Image.DecompressionBombError: Image size (256901120 pixels) exceeds limit of 178956970 pixels, could be decompression bomb DOS attack..
I also try to add the demo of Image.MAX_IMAGE_PIXELS = None behind all demo of from PIL import Image because I can't find it in cvat\apps\engine\task.py,but I got annother error
Could not create the task Error: Work-horse process was terminated unexpectedly (waitpid returned 9).
is three another method?
Hi, I generated white empty image in Pillow with such the size and was able to create a task after adding
from PIL import Image
Image.MAX_IMAGE_PIXELS = None
to cvat/settings/base.py
I do as you talked,when I click the submit,it seems successfuly implement The data is being uploaded to the server..,but I get another error in the step of Media files are being extracted...
Could not create the task Error: Could not create the task on the server. Work-horse process was terminated unexpectedly (waitpid returned 9).
the image I submit is 490MB(17920*14400),it seems the process wait a long time and terminated.
cvat_proxy | 172.18.0.1 - - [22/Sep/2020:08:54:44 +0000] "GET /api/v1/tasks/68/status HTTP/1.1" 200 66 "http://localhost:8080/tasks/create" "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/85.0.4183.102 Safari/537.36" cvat_db | 2020-09-22 08:55:04.547 UTC [41] LOG: unexpected EOF on client connection with an open transaction cvat | 2020-09-22 08:55:06,703 DEBG 'rqworker_default_1' stderr output: cvat | WARNING:rq.worker:Moving job to FailedJobRegistry (work-horse terminated unexpectedly; waitpid returned 9) cvat |
I can suggest you try to install CVAT with connected shared storage. You can create tasks browsing the storage client side and selecting necessary files.
@tingfengwanxiao , I don't think that we support of such big images for now. Need to slice it to show efficiently in UI.
It looks like the process has crashed due to out of memory error. You can check it with dmesg -HTw
on the host.
@nmanovic Hi any update on this issue?
I try to upload large images (800M Pixels) and receive this error:
I am modifying these:
supervisord:
base.py:
manage.py:
To modify these I am running docker-compose stop and then changing these files and parameters and then running docker-compose up -d.
- Is this the correct way to effectively modify files in the container
- Is it possible to upload large images in docker e.g 800M Pixels
Thank you! Sam @zhiltsov-max
@nmanovic Hi any update on this issue?
I try to upload large images (800M Pixels) and receive this error:
I am modifying these:
supervisord:
base.py:
manage.py:
To modify these I am running docker-compose stop and then changing these files and parameters and then running docker-compose up -d.
- Is this the correct way to effectively modify files in the container
- Is it possible to upload large images in docker e.g 800M Pixels
Thank you! Sam @zhiltsov-max
Hey did it work?
It did not worked 🥲
Also i tried this, it did not worked.
from PIL import Image
Image.MAX_IMAGE_PIXELS = 100000000000
So last option i have create --> Smaller Sets of Task like 50 images ⭐️
@gamingflexer We were able to upload huge images e.g 4000 x 8000 by mounting a share drive to the docker container - it's in CVAT documentation. And the task upload is instant aswell. Cheers
@gamingflexer We were able to upload huge images e.g 4000 x 8000 by mounting a share drive to the docker container - it's in CVAT documentation. And the task upload is instant aswell. Cheers
Thanks ! i will try that for sure
@gamingflexer We were able to upload huge images e.g 4000 x 8000 by mounting a share drive to the docker container - it's in CVAT documentation. And the task upload is instant aswell. Cheers
Thanks ! i will try that for sure
The issue still persists i have mounted the docker external drive and it is visible in cvat but still same.

Hello,
Is there any update on this? I was able to upload images to the task but I get "Server Error 500" when attempting to open up the task itself.
I uploaded the image by adding the line "Image.MAX_IMAGE_PIXELS = None" to my import workers and loading the image from connected file share.
The image is approximately 18000 x 36000 pixels, ~150 MB.
If you attach cloud storage, you can use bigger images than PIL limit. (https://docs.cvat.ai/docs/manual/basics/attach-cloud-storage/)
I can upload an image with size (20000 x 14000), but there is another error throw out:
DOMException: CanvasRenderingContext2D.scale: Canvas exceeds max size.
Do you have any idea on this? It may come from browser support and really hard to fix https://jhildenbiddle.github.io/canvas-size/#/?id=test-results