Case size mismatch

Dear organizers,

I find the cases have different sizes with Groundtruth. The cases are listed below. lnq2023-train-0418 lnq2023-train-0320 lnq2023-train-0197 lnq2023-train-0394 lnq2023-train-0618 lnq2023-train-1047 lnq2023-train-0658 lnq2023-train-0225 lnq2023-train-0934 lnq2023-train-0159 lnq2023-train-0034 lnq2023-train-0720 lnq2023-train-0977 lnq2023-train-0125 lnq2023-train-0655 lnq2023-train-0995 lnq2023-train-0402 lnq2023-train-0551 lnq2023-train-0668 lnq2023-train-1081 lnq2023-train-0009 lnq2023-train-0793 lnq2023-train-0874 lnq2023-train-0279 lnq2023-train-0283 lnq2023-train-0752 lnq2023-train-0798 lnq2023-train-0656 lnq2023-train-0150 lnq2023-train-0302 lnq2023-train-0909 lnq2023-train-1044 lnq2023-train-0183 lnq2023-train-0504 lnq2023-train-0328 lnq2023-train-0610 lnq2023-train-0727 lnq2023-train-0826 lnq2023-train-0992

Best regards, Eugene

Hello,

@organizers I would also be interested in why the image ground truth pairs don't have identical shapes? Is there a reason for that or if not, is there a script to fix the misalignment?

@eugene did you get a reply from the organisers regarding this?

Best regards, Tassilo

The size discrepancy resulted from a bug in the pre-processing step. Despite this disparity, the positioning of the segmentation labels was accurate. We have fixed the issue and updated the Training Data with the corrected labels. If you only need corrected labels, you can find the link on the Dataset page under the "Corrected Labels" section.

Thanks a lot for the quick fix.

Cheers, Tassilo