Upon reviewing the dataset, I found that while most input data has a shape of (320, 320, 320) as boxel data, approximately 5% contains data with shape (256, 256, 256). Additionally, one instance was identified with shape (384, 384, 384) (ID: 3866421231).
\nI'd like to verify the origin of these data points. Are these instances created by resizing (320, 320, 320) data to (256, 256, 256) while maintaining equivalent range, or were they cropped to (256, 256, 256) size because some portions fell outside the scan area?\nI'd also like to confirm the same regarding the data with shape (384, 384, 384).
\nFurthermore, will data in this format also be included in the test dataset for submission?
", "messages": [ { "id": "3345022", "postDate": "11/23/2025 06:47:53", "content": "Upon reviewing the dataset, I found that while most input data has a shape of (320, 320, 320) as boxel data, approximately 5% contains data with shape (256, 256, 256). Additionally, one instance was identified with shape (384, 384, 384) (ID: 3866421231).
\nI'd like to verify the origin of these data points. Are these instances created by resizing (320, 320, 320) data to (256, 256, 256) while maintaining equivalent range, or were they cropped to (256, 256, 256) size because some portions fell outside the scan area?\nI'd also like to confirm the same regarding the data with shape (384, 384, 384).
\nFurthermore, will data in this format also be included in the test dataset for submission?
", "rawMarkdown": "Upon reviewing the dataset, I found that while most input data has a shape of (320, 320, 320) as boxel data, approximately 5% contains data with shape (256, 256, 256). Additionally, one instance was identified with shape (384, 384, 384) (ID: 3866421231).\n\nI'd like to verify the origin of these data points. Are these instances created by resizing (320, 320, 320) data to (256, 256, 256) while maintaining equivalent range, or were they cropped to (256, 256, 256) size because some portions fell outside the scan area?\nI'd also like to confirm the same regarding the data with shape (384, 384, 384).\n\nFurthermore, will data in this format also be included in the test dataset for submission?", "votes": null }, { "id": "3346214", "postDate": "11/24/2025 07:20:47", "content": "Thank you for the question!
\nThe full scan volume is of the order of 10 thousands voxels in each direction.\nThe dataset for the competition is made of chunks that are smaller annotated crops of this dataset.\nSince we don't know which is the optimal 3D patch size, we provided chunks of different dimension. The dimension choice depends by our capability of annotating that specific region of interest.\nThe chunks have not been resized.
", "rawMarkdown": "Thank you for the question!\n\nThe full scan volume is of the order of 10 thousands voxels in each direction.\nThe dataset for the competition is made of chunks that are smaller annotated crops of this dataset.\nSince we don't know which is the optimal 3D patch size, we provided chunks of different dimension. The dimension choice depends by our capability of annotating that specific region of interest.\nThe chunks have not been resized.", "votes": null }, { "id": "3348791", "postDate": "11/26/2025 07:40:34", "content": "The train set has cubes with lengths 256, 320, 384. Are the test set cubes limited to these 3 lengths as well?
", "rawMarkdown": "The train set has cubes with lengths 256, 320, 384. Are the test set cubes limited to these 3 lengths as well?", "votes": null }, { "id": "3349709", "postDate": "11/26/2025 22:53:31", "content": "Hello! Yes they are , although i thought i had restricted to only 256 and 384 (with 256 representing a small fraction of the total). I'll have to take another look through
", "rawMarkdown": "Hello! Yes they are , although i thought i had restricted to only 256 and 384 (with 256 representing a small fraction of the total). I'll have to take another look through", "votes": null }, { "id": "3361333", "postDate": "12/03/2025 09:49:02", "content": "Could you confirm this?\nI'm having issues with my submission even though it runs fine on the train set and I suspect it's because of cubes larger than 384 in the test set.
", "rawMarkdown": "Could you confirm this?\nI'm having issues with my submission even though it runs fine on the train set and I suspect it's because of cubes larger than 384 in the test set.", "votes": null } ], "comments": [ { "id": 3346214, "author_name": "giorgioangelotti", "author_url": "", "post_date": "11/24/2025 07:20:47", "content": "Thank you for the question!
\nThe full scan volume is of the order of 10 thousands voxels in each direction.\nThe dataset for the competition is made of chunks that are smaller annotated crops of this dataset.\nSince we don't know which is the optimal 3D patch size, we provided chunks of different dimension. The dimension choice depends by our capability of annotating that specific region of interest.\nThe chunks have not been resized.
", "votes": null, "replies": [ { "id": 3348791, "author_name": "sroger", "author_url": "", "post_date": "11/26/2025 07:40:34", "content": "The train set has cubes with lengths 256, 320, 384. Are the test set cubes limited to these 3 lengths as well?
", "votes": null, "replies": [ { "id": 3349709, "author_name": "seanjohnsonsp", "author_url": "", "post_date": "11/26/2025 22:53:31", "content": "Hello! Yes they are , although i thought i had restricted to only 256 and 384 (with 256 representing a small fraction of the total). I'll have to take another look through
", "votes": null, "replies": [ { "id": 3361333, "author_name": "sroger", "author_url": "", "post_date": "12/03/2025 09:49:02", "content": "Could you confirm this?\nI'm having issues with my submission even though it runs fine on the train set and I suspect it's because of cubes larger than 384 in the test set.
", "votes": null, "replies": [] } ] } ] } ] } ], "raw_markdown_by_id": { "3345022": "Upon reviewing the dataset, I found that while most input data has a shape of (320, 320, 320) as boxel data, approximately 5% contains data with shape (256, 256, 256). Additionally, one instance was identified with shape (384, 384, 384) (ID: 3866421231).\n\nI'd like to verify the origin of these data points. Are these instances created by resizing (320, 320, 320) data to (256, 256, 256) while maintaining equivalent range, or were they cropped to (256, 256, 256) size because some portions fell outside the scan area?\nI'd also like to confirm the same regarding the data with shape (384, 384, 384).\n\nFurthermore, will data in this format also be included in the test dataset for submission?", "3346214": "Thank you for the question!\n\nThe full scan volume is of the order of 10 thousands voxels in each direction.\nThe dataset for the competition is made of chunks that are smaller annotated crops of this dataset.\nSince we don't know which is the optimal 3D patch size, we provided chunks of different dimension. The dimension choice depends by our capability of annotating that specific region of interest.\nThe chunks have not been resized.", "3348791": "The train set has cubes with lengths 256, 320, 384. Are the test set cubes limited to these 3 lengths as well?", "3349709": "Hello! Yes they are , although i thought i had restricted to only 256 and 384 (with 256 representing a small fraction of the total). I'll have to take another look through", "3361333": "Could you confirm this?\nI'm having issues with my submission even though it runs fine on the train set and I suspect it's because of cubes larger than 384 in the test set." }, "source": "meta" }