-
Notifications
You must be signed in to change notification settings - Fork 28
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
Warnings + time-out when sending a job to deepcell in Segment_image_data #297
Comments
How many images are you trying to run at once? And how large are they? |
6 images, 2048x2048, 31 channels |
Can you try sending a single 1024x1024 crop of one of the images? That doesn't seem like it would be too large, but we haven't tried larger than 1024 so far |
Resizing to 1024x1024 completes the call. Unfortunately, the deepcell results zip file is empty (22 bytes). |
I tried again and got a json report (didn't get that last time):
|
And the same zip file works on the website? |
Good question -- is there an easy way to stop create_deepcell_output from deleting the input zip file? |
Yeah we should probably remove that line anyway. In the meantime you can copy the function contents into a cell and edit it:
|
I was mistaken (found a typo in my script)! It works at 1024x1024 resolution. So I guess the original problem was just that it was too big. Is there any plan to allow queries of images that are bigger than 1024? |
What was the typo that lead to an empty zip file being returned? Ideally we'd like to catch that type of thing earlier. To clarify, you took a 2048x2048 image with 31 channels, picked 2 of them for the nuclear/segmentation channel, and that image failed with the timeout error? |
My image ended in "MassCorrected-Filtered-1.tiff" and my MIBItiff_suffix variable = "MassCorrected-Filtered.tiff" without the '-1'. Yes, except we are using 2 nuclear channels and 3 membrane channels. And this works at 2048x2048 if we zip manually and upload through the browser. |
Got it. The wrong formatting for input files will be addressed by #281. |
I realized I had never tried a single 2048x2048 image, and it actually worked through the create_deepcell_output. So just to re-cap: ✓ Single 1024x1024 via both deepcell.org & create_deepcell_output method If I zip the 6 files manually (on Linux), the resulting zip file is 30.2MB. The one generated by the jupyter notebook is 201.3MB. |
Okay, so there's an issue with our zip file creation which is making comically large zip files. This is triggering a known bug in the kiosk-client which times out for large uploads. We'll address the file size issue first. We know the limit is somewhere around 100 MB, so once we fix the zip issue, that should give us around 75 1024x1024 before running into issues |
Great! So, just going by # of pixels, that's about 20 2048x2048 images right? Or would it scale differently? |
Normally I would say yes, but whatever is creating the massive zip files could be leading to creative, non-linear scaling as well. The newest version of master branch no longer deletes the zip files, so you can always manually upload it to deepcell.org if it pops up again. |
I'm trying to send run the create_deepcell_output method to zip and send a file to the DeepCell server (very minor: there's a typo in the uploading message that says DeppCell, which is kinda great).
I get several warnings and then it just never completes the send and seems to time out. When I kill the process, the .zip file gets removed.
Thanks for the help!
What have you tried so far?
To get around it I've been just zipping myself and uploading it through the browser. Haven't been able to figure out it out in the code though.
Additional context
Here's the error trace in case that's helpful
The text was updated successfully, but these errors were encountered: