The algorithm failed on one or more cases.

I met a problem that my algorithm failed but I have no log about the how to fix it. So could u provide the log to us for solving this problem.

Your error log has been shared with you; please check your inbox of Grand-Challenge account.

The following suggestions may help you:

1. You don't need to create the input and output folders when preparing the Docker image for submission. Instead, follow the guidelines and directory structure outlined here. Additionally, do not use the inference file you use for local testing. Instead, use the inference file provided here.

2. If you encounter GPU memory issues when using sliding window inference, try transferring the data to the CPU instead of the GPU. You can do this as shown below:

   sliding_window_inference(image, (patch_size, patch_size, patch_size), num_samples, model, device='cpu')

I hope this resolves the issue.

Thank you for reply. I have followed the code from the inference.py from the GitHub link and changed the device to ‘cpu’, but it failed again. Could you please tell me the log of error? Thank you so much.

Your error log has been shared with you; please check your inbox of Grand-Challenge account.

Sorry for failed submission again. According to last error log, I changed the data type of array and permuted the dimension. Moreover, I ran the test.sh file and it successed. So could you tell me what I can do for successful submission? And please send me the error log again. Thank you.

The error log indicates that your algorithm exceeded the time limit. Please ensure that it can generate the segmentation files within the five-minute limit.

Hello, I have made several attempts in the past few days.

Firstly, regarding the issue with the output data type, I ensured that the saved image data type is uint8 and the dimensions are correct.

Next, I addressed the timeout issue with four attempts: 1. I switched the inference device to GPU. 2. I directly used the official inference.py script without loading weights. 3. I generated an image with dimensions matching the input image, using uint8 data type, which took 0.5 seconds locally. 4. I saw on a forum that someone mentioned the Docker image being too large, so I tried a smaller Docker image.

All of these attempts have failed. Could you please advise on specific modifications I should make to ensure a successful submission?