{ "id": 459470, "title": "Question about rle", "url": "/competitions/blood-vessel-segmentation/discussion/459470", "author_name": "", "post_date": "2023-12-05T10:01:32.331864700Z", "votes": 1, "comment_count": 4, "views": 0, "content": "

For the rle, I could not fully understand what is rle, anyone who can explain the rle results and how to score the rle results, thanks a lot

", "messages": [ { "id": "2549506", "postDate": "12/05/2023 10:01:32", "content": "

For the rle, I could not fully understand what is rle, anyone who can explain the rle results and how to score the rle results, thanks a lot

", "rawMarkdown": "For the rle, I could not fully understand what is rle, anyone who can explain the rle results and how to score the rle results, thanks a lot", "votes": null }, { "id": "2550421", "postDate": "12/06/2023 02:58:38", "content": "

As the overview said:>For each id in the test set, you must predict rle, a run-length encoded instance segmentation mask, where id represents {dataset}_{slice} for an image with path test/{dataset}/images/{slice}.tif. Represent the RLE for an empty mask as 1 0.
\nFor myself,you can not upload your predict as image,rle is a type you can upload and scored easily.You can also find the function of rle_encode from https://www.kaggle.com/code/metric/surface-dice-metric/notebook.Good luck.

", "rawMarkdown": "As the overview said:>For each id in the test set, you must predict rle, a run-length encoded instance segmentation mask, where id represents {dataset}_{slice} for an image with path test/{dataset}/images/{slice}.tif. Represent the RLE for an empty mask as 1 0.\nFor myself,you can not upload your predict as image,rle is a type you can upload and scored easily.You can also find the function of rle_encode from [https://www.kaggle.com/code/metric/surface-dice-metric/notebook](url).Good luck.", "votes": null }, { "id": "2552130", "postDate": "12/07/2023 08:07:04", "content": "

Thanks a lot

", "rawMarkdown": "Thanks a lot", "votes": null }, { "id": "2552679", "postDate": "12/07/2023 16:44:17", "content": "

@johnzjz check this discussion

", "rawMarkdown": "johnzjz check [this](https://www.kaggle.com/competitions/vesuvius-challenge-ink-detection/discussion/395067) discussion", "votes": null }, { "id": "2555477", "postDate": "12/10/2023 00:23:24", "content": "

I am still trying to wrap my mind around it as well. According to the code any section of a label image that has consecutive positive values. Since it is a 1D array, I am assuming that if an object is bigger in one dimension there would be consecutive lengths for each set of pixels on the other axis.

", "rawMarkdown": "I am still trying to wrap my mind around it as well. According to the code any section of a label image that has consecutive positive values. Since it is a 1D array, I am assuming that if an object is bigger in one dimension there would be consecutive lengths for each set of pixels on the other axis.", "votes": null } ], "comments": [ { "id": 2550421, "author_name": "linenkai", "author_url": "", "post_date": "12/06/2023 02:58:38", "content": "

As the overview said:>For each id in the test set, you must predict rle, a run-length encoded instance segmentation mask, where id represents {dataset}_{slice} for an image with path test/{dataset}/images/{slice}.tif. Represent the RLE for an empty mask as 1 0.
\nFor myself,you can not upload your predict as image,rle is a type you can upload and scored easily.You can also find the function of rle_encode from https://www.kaggle.com/code/metric/surface-dice-metric/notebook.Good luck.

", "votes": null, "replies": [ { "id": 2552130, "author_name": "johnzjz", "author_url": "", "post_date": "12/07/2023 08:07:04", "content": "

Thanks a lot

", "votes": null, "replies": [] } ] }, { "id": 2552679, "author_name": "alejopaullier", "author_url": "", "post_date": "12/07/2023 16:44:17", "content": "

@johnzjz check this discussion

", "votes": null, "replies": [] }, { "id": 2555477, "author_name": "tuttlen", "author_url": "", "post_date": "12/10/2023 00:23:24", "content": "

I am still trying to wrap my mind around it as well. According to the code any section of a label image that has consecutive positive values. Since it is a 1D array, I am assuming that if an object is bigger in one dimension there would be consecutive lengths for each set of pixels on the other axis.

", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "2549506": "For the rle, I could not fully understand what is rle, anyone who can explain the rle results and how to score the rle results, thanks a lot", "2550421": "As the overview said:>For each id in the test set, you must predict rle, a run-length encoded instance segmentation mask, where id represents {dataset}_{slice} for an image with path test/{dataset}/images/{slice}.tif. Represent the RLE for an empty mask as 1 0.\nFor myself,you can not upload your predict as image,rle is a type you can upload and scored easily.You can also find the function of rle_encode from [https://www.kaggle.com/code/metric/surface-dice-metric/notebook](url).Good luck.", "2552130": "Thanks a lot", "2552679": "johnzjz check [this](https://www.kaggle.com/competitions/vesuvius-challenge-ink-detection/discussion/395067) discussion", "2555477": "I am still trying to wrap my mind around it as well. According to the code any section of a label image that has consecutive positive values. Since it is a 1D array, I am assuming that if an object is bigger in one dimension there would be consecutive lengths for each set of pixels on the other axis." }, "source": "meta" }