The algorithm failed on one or more cases-final_submit

The algorithm failed on one or more cases-final_submit  

  By: AI_medical on Aug. 29, 2022, 1:25 a.m.

Dear Marc, The algorithm I finally submitted is the same as the one I tested initially, and this error was reported. Could you give me some information? Or is it because the machine is overloaded and unsuccessful?

Thanks Wang

Re: The algorithm failed on one or more cases-final_submit  

  By: ekbreininger on Aug. 29, 2022, 7:14 a.m.

Hi Wang,

Your algorithm exceeded the time limit on several images. If this exact container worked on the preliminary test set before reliably, kindly try to resubmit, potentially the servers were busy.

Thanks, Katharina

Re: The algorithm failed on one or more cases-final_submit  

  By: AI_medical on Aug. 29, 2022, 7:27 a.m.

Dear Katharina, Yes, I have been running successfully on the preliminary test set, can you help me to run it again, it shows me 'A submission for this algorithm container image for this phase already exists'.

Thanks Wang

Re: The algorithm failed on one or more cases-final_submit  

  By: ekbreininger on Aug. 29, 2022, 8:42 a.m.

Hi Wang,

This is likely again because of the issue mentioned in a previous post (and with the same solution): https://grand-challenge.org/forums/forum/mitosis-domain-generalization-challenge-2022-671/topic/a-submission-for-this-algorithm-container-image-for-this-phase-already-exists-1148/

Could you kindly try to do this? I tried to request access to your algorithm to re-submit for you, but we will likely get the same error.

Best, Katharina

Re: The algorithm failed on one or more cases-final_submit  

  By: AI_medical on Aug. 29, 2022, 11:38 a.m.

Dear, I tried again, but still reported this error, I had been successful in the preliminarytest set.Can you check it out? what is the reason? Thanks Wang

Re: The algorithm failed on one or more cases-final_submit  

  By: ekbreininger on Aug. 29, 2022, 2:14 p.m.

Hi Wang,

Note: I just saw that the current run of your algorithm seems to pass for the images that is previously crashed on. So kindly treat the comment below as a hint:

For the previous run of your container: I see the same issue as before: Your algorithm runs out of time for a subset of images. Specifically, it doesn't even start the image loop within the time limit of 20 mins. In general, your container is also on the successful ones close-ish to the 20min mark. One issue that I have found: You split the image into 130-160 patches for some cases, but into 207 for others. The ones with 207 seem to be the ones that are failing.

If the current submission does not run successfully afterall: Could you check whether there could be an issue there?

Best, Katharina