Question about the usage of the BlueCat dataset in the released train/val/test splits

[huangxintheone]

Hello, thank you very much for sharing your excellent work and open-sourcing the code.

While reading the paper, I noticed that the BlueCat dataset is listed as one of the datasets used in the experiments. However, when I examined the released source code—specifically the predefined train, val, and test split lists—I could not find any entries corresponding to the BlueCat dataset.

I would like to ask for clarification on this point:

Was the BlueCat dataset used only in certain experiments (e.g., ablation studies or supplementary experiments) that are not fully reflected in the released code?

Or was BlueCat merged or renamed under another dataset identifier in the data split lists?

Alternatively, is the BlueCat-related data split not included in the current public release?

Understanding this would be very helpful for accurately reproducing the experimental setup described in the paper.

Thank you again for your time and for making the code available.
Looking forward to your clarification.

Best regards

[bxiang233]

Hello, thank you very much for sharing your excellent work and open-sourcing the code.

While reading the paper, I noticed that the BlueCat dataset is listed as one of the datasets used in the experiments. However, when I examined the released source code—specifically the predefined train, val, and test split lists—I could not find any entries corresponding to the BlueCat dataset.

I would like to ask for clarification on this point:

Was the BlueCat dataset used only in certain experiments (e.g., ablation studies or supplementary experiments) that are not fully reflected in the released code?

Or was BlueCat merged or renamed under another dataset identifier in the data split lists?

Alternatively, is the BlueCat-related data split not included in the current public release?

Understanding this would be very helpful for accurately reproducing the experimental setup described in the paper.

Thank you again for your time and for making the code available. Looking forward to your clarification.

Best regards

Hello, thanks for your question, and sorry for the confusion.

BlueCat was used in the experiments reported in the paper and was handled in the same way as the other regions. Its absence in the current main branch is due to changes introduced after commit 28d8045 (see commit 28d8045, “Add batch loader script and update train/test lists”), where atakan fixed several issues and updated train/val/test lists.

If you check the earlier version before this commit, you will find that BlueCat appears in the train_list, val_list, and test_list, and is treated identically to the other datasets.

[barkhaneum]

Hello, thank you very much for sharing your excellent work and open-sourcing the code.
While reading the paper, I noticed that the BlueCat dataset is listed as one of the datasets used in the experiments. However, when I examined the released source code—specifically the predefined train, val, and test split lists—I could not find any entries corresponding to the BlueCat dataset.
I would like to ask for clarification on this point:
Was the BlueCat dataset used only in certain experiments (e.g., ablation studies or supplementary experiments) that are not fully reflected in the released code?
Or was BlueCat merged or renamed under another dataset identifier in the data split lists?
Alternatively, is the BlueCat-related data split not included in the current public release?
Understanding this would be very helpful for accurately reproducing the experimental setup described in the paper.
Thank you again for your time and for making the code available. Looking forward to your clarification.
Best regards

Hello, thanks for your question, and sorry for the confusion.

BlueCat was used in the experiments reported in the paper and was handled in the same way as the other regions. Its absence in the current main branch is due to changes introduced after commit 28d8045 (see commit 28d8045, “Add batch loader script and update train/test lists”), where atakan fixed several issues and updated train/val/test lists.

If you check the earlier version before this commit, you will find that BlueCat appears in the train_list, val_list, and test_list, and is treated identically to the other datasets.

Hello. Thank you for sharing this excellent paper and dataset. I have an additional question.

The published BlueCat dataset appears to be a merged, complete dataset.

Do I need to tiling the dataset before training?

Or will training work if I follow the instructions in the git repository?

[bxiang233]

Hello, thank you very much for sharing your excellent work and open-sourcing the code.
While reading the paper, I noticed that the BlueCat dataset is listed as one of the datasets used in the experiments. However, when I examined the released source code—specifically the predefined train, val, and test split lists—I could not find any entries corresponding to the BlueCat dataset.
I would like to ask for clarification on this point:
Was the BlueCat dataset used only in certain experiments (e.g., ablation studies or supplementary experiments) that are not fully reflected in the released code?
Or was BlueCat merged or renamed under another dataset identifier in the data split lists?
Alternatively, is the BlueCat-related data split not included in the current public release?
Understanding this would be very helpful for accurately reproducing the experimental setup described in the paper.
Thank you again for your time and for making the code available. Looking forward to your clarification.
Best regards

Hello, thanks for your question, and sorry for the confusion.
BlueCat was used in the experiments reported in the paper and was handled in the same way as the other regions. Its absence in the current main branch is due to changes introduced after commit 28d8045 (see commit 28d8045, “Add batch loader script and update train/test lists”), where atakan fixed several issues and updated train/val/test lists.
If you check the earlier version before this commit, you will find that BlueCat appears in the train_list, val_list, and test_list, and is treated identically to the other datasets.

Hello. Thank you for sharing this excellent paper and dataset. I have an additional question.

The published BlueCat dataset appears to be a merged, complete dataset.

Do I need to tiling the dataset before training?

Or will training work if I follow the instructions in the git repository?

Hi, you do not need to tile the dataset before training. Training should work if you follow the README step by step.

Just one note: in our experiments, we used A100 GPUs. I am not entirely sure how it will behave on other GPUs in terms of memory usage. Feel free to try it and see how it goes.