Output Format for Algorithm Results

Hello everyone,

we need to clarify the output format for the algorithm results.

Could you please advise on the following:

• What data type and value range (0-1, 0-255) should the output image of our algorithm return?

• Should the output be saved as an image file (e.g., .png), or is it sufficient to return it as a PyTorch/TensorFlow tensor or a NumPy array?

Additionally, it would be extremely helpful if you could upload a short example code snippet demonstrating the exact expected output format. This will ensure everyone understands how to submit their results for evaluation.

Ideally, we would also appreciate a brief example script for the metric evaluation code itself, so we can see how the outputs will be assessed.

Regards,

Dominik

Hi Dominik,

Great questions. The pixel values should be in the 0-255 range (uint8).

We've actually set up a GitHub repo with a complete example to show exactly how this works. It should answer all your questions. Link: https://github.com/yunshuli2002/DehazingEcho2025

Hope that helps! Let us know if anything is still unclear after checking the repo.

Best, Xue Gao

Hi Xue,

Thank you so much for the repo link! It is really helpful!

Could you please share your score-evaluation-functions also? We are not entirely sure how they work under the hood. It would be great if you could provide a short example script for the metric computation so that we can implement it correctly in our own scripts, independent of the docker container.

Thanks in advance and best regards

Dominik

Hi Dominik,

We appreciate your suggestion and have now updated the GitHub repository to include the score-evaluation-functions as well. Please feel free to check the link for the latest version.

Best regards, Xue Gao