{
  "id": 285321,
  "title": "What does it mean by: \" LIVECell is the predecessor dataset to this competition\" ?",
  "url": "/competitions/sartorius-cell-instance-segmentation/discussion/285321",
  "author_name": "",
  "post_date": "2021-11-04T07:20:17.990798Z",
  "votes": 14,
  "comment_count": 10,
  "views": 0,
  "content": "<p>Does it mean that there is no overlapping images between LIVECell dataset and the competition dataset in term of SH-SY5Y cell line ? <a href=\"https://www.kaggle.com/christoffersartorius\" target=\"_blank\">@christoffersartorius</a> </p>",
  "messages": [
    {
      "id": "1570484",
      "postDate": "11/04/2021 07:20:17",
      "content": "<p>Does it mean that there is no overlapping images between LIVECell dataset and the competition dataset in term of SH-SY5Y cell line ? <a href=\"https://www.kaggle.com/christoffersartorius\" target=\"_blank\">@christoffersartorius</a> </p>",
      "rawMarkdown": "Does it mean that there is no overlapping images between LIVECell dataset and the competition dataset in term of SH-SY5Y cell line ? @christoffersartorius",
      "votes": null
    },
    {
      "id": "1570825",
      "postDate": "11/04/2021 12:04:24",
      "content": "<p>I am not sure if I get it right.   in some discussion, It states that list of dictionary vs dictionary of dictionary.</p>",
      "rawMarkdown": "I am not sure if I get it right.   in some discussion, It states that list of dictionary vs dictionary of dictionary.",
      "votes": null
    },
    {
      "id": "1570869",
      "postDate": "11/04/2021 12:48:48",
      "content": "<p>Hi The Nam.</p>\n<p>LIVECell is a publication with we did in nature methods where we presented a resource with 1.6 million annotated cells. From our work with that dataset we noticed that our models struggle with neuronal cells, more specifically the SH-SY5Y.</p>\n<p>The dataset presented in this competition is a product of a second round of gathering data and annotating with the specific purpose of generating stronger models with regards to neuronal cells. The SH-SY5Y images from the two datasets should not overlap.</p>\n<p>I hope that answers your questions.</p>\n<p>Kindly,<br>\nChristoffer</p>",
      "rawMarkdown": "Hi The Nam.\n\nLIVECell is a publication with we did in nature methods where we presented a resource with 1.6 million annotated cells. From our work with that dataset we noticed that our models struggle with neuronal cells, more specifically the SH-SY5Y.\n\nThe dataset presented in this competition is a product of a second round of gathering data and annotating with the specific purpose of generating stronger models with regards to neuronal cells. The SH-SY5Y images from the two datasets should not overlap.\n\nI hope that answers your questions.\n\nKindly,\nChristoffer",
      "votes": null
    },
    {
      "id": "1570935",
      "postDate": "11/04/2021 13:46:49",
      "content": "<p>Thanks for your reply, I understand now.</p>",
      "rawMarkdown": "Thanks for your reply, I understand now.",
      "votes": null
    },
    {
      "id": "1570948",
      "postDate": "11/04/2021 13:55:36",
      "content": "<blockquote>\n  <p>LIVECell is a publication with we did in <strong>nature methods</strong></p>\n</blockquote>\n<p>Is the dataset in this competition annotated using the same method as describe in the LiveCell paper? If there is any difference, can we know about that?</p>",
      "rawMarkdown": "> LIVECell is a publication with we did in **nature methods**\n\nIs the dataset in this competition annotated using the same method as describe in the LiveCell paper? If there is any difference, can we know about that?",
      "votes": null
    },
    {
      "id": "1570971",
      "postDate": "11/04/2021 14:18:26",
      "content": "<p>FYI, I've gathered things I've found about the dataset here <a href=\"https://www.kaggle.com/c/sartorius-cell-instance-segmentation/discussion/285384\" target=\"_blank\">https://www.kaggle.com/c/sartorius-cell-instance-segmentation/discussion/285384</a></p>",
      "rawMarkdown": "FYI, I've gathered things I've found about the dataset here https://www.kaggle.com/c/sartorius-cell-instance-segmentation/discussion/285384",
      "votes": null
    },
    {
      "id": "1571581",
      "postDate": "11/05/2021 01:34:54",
      "content": "<p><a href=\"https://www.kaggle.com/christoffersartorius\" target=\"_blank\">@christoffersartorius</a> Can you please answer the above question ? </p>",
      "rawMarkdown": "christoffersartorius Can you please answer the above question ?",
      "votes": null
    },
    {
      "id": "1572159",
      "postDate": "11/05/2021 13:12:29",
      "content": "<p><a href=\"https://www.kaggle.com/christoffersartorius\" target=\"_blank\">@christoffersartorius</a> </p>\n<p>I have a similar question.<br>\nAre LIVECell and this competiton same annotator and same process right?<br>\nIt's important for building processing.</p>",
      "rawMarkdown": "christoffersartorius \n\nI have a similar question.\nAre LIVECell and this competiton same annotator and same process right?\nIt's important for building processing.",
      "votes": null
    },
    {
      "id": "1572222",
      "postDate": "11/05/2021 14:25:31",
      "content": "<p>In my observation, for LiveCell dataset, consider only Shsy5y images, there are images with about 1400 instances, which is much higher than that of the competition data. Another point is Shsy5y  LiveCell data has some images looking alike Astro and Cort. </p>",
      "rawMarkdown": "In my observation, for LiveCell dataset, consider only Shsy5y images, there are images with about 1400 instances, which is much higher than that of the competition data. Another point is Shsy5y  LiveCell data has some images looking alike Astro and Cort.",
      "votes": null
    },
    {
      "id": "1572350",
      "postDate": "11/05/2021 15:52:24",
      "content": "<p>thanks for your replay.<br>\nshsy5y is maybe good external dataset than other cells.  <br>\nI found same question from you in thread. </p>\n<p>I'm looking forward to answer from host.</p>",
      "rawMarkdown": "thanks for your replay.\nshsy5y is maybe good external dataset than other cells.  \nI found same question from you in thread. \n\nI'm looking forward to answer from host.",
      "votes": null
    },
    {
      "id": "1572481",
      "postDate": "11/05/2021 17:47:59",
      "content": "<blockquote>\n  <p>Another point is Shsy5y LiveCell data has some images looking alike Astro and Cort.</p>\n</blockquote>\n<p>Indeed! Here is an example image from the LiveCell dataset. This image (SHSY5Y_Phase_C10_2_00d04h00m_3.tif) is from the livecell_train_val_images/SHSY5Y directory, but does not look anything like SHSY5Y.</p>\n<p><img src=\"https://raw.githubusercontent.com/anlthms/image-repo/main/SHSY5Y_Phase_C10_2_00d04h00m_3.png\" alt=\"\"></p>",
      "rawMarkdown": "> Another point is Shsy5y LiveCell data has some images looking alike Astro and Cort.\n\nIndeed! Here is an example image from the LiveCell dataset. This image (SHSY5Y_Phase_C10_2_00d04h00m_3.tif) is from the livecell_train_val_images/SHSY5Y directory, but does not look anything like SHSY5Y.\n\n![](https://raw.githubusercontent.com/anlthms/image-repo/main/SHSY5Y_Phase_C10_2_00d04h00m_3.png)",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1570825,
      "author_name": "dragonzhang",
      "author_url": "",
      "post_date": "11/04/2021 12:04:24",
      "content": "<p>I am not sure if I get it right.   in some discussion, It states that list of dictionary vs dictionary of dictionary.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 1570869,
      "author_name": "christoffersartorius",
      "author_url": "",
      "post_date": "11/04/2021 12:48:48",
      "content": "<p>Hi The Nam.</p>\n<p>LIVECell is a publication with we did in nature methods where we presented a resource with 1.6 million annotated cells. From our work with that dataset we noticed that our models struggle with neuronal cells, more specifically the SH-SY5Y.</p>\n<p>The dataset presented in this competition is a product of a second round of gathering data and annotating with the specific purpose of generating stronger models with regards to neuronal cells. The SH-SY5Y images from the two datasets should not overlap.</p>\n<p>I hope that answers your questions.</p>\n<p>Kindly,<br>\nChristoffer</p>",
      "votes": null,
      "replies": [
        {
          "id": 1570935,
          "author_name": "namgalielei",
          "author_url": "",
          "post_date": "11/04/2021 13:46:49",
          "content": "<p>Thanks for your reply, I understand now.</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 1570948,
          "author_name": "namgalielei",
          "author_url": "",
          "post_date": "11/04/2021 13:55:36",
          "content": "<blockquote>\n  <p>LIVECell is a publication with we did in <strong>nature methods</strong></p>\n</blockquote>\n<p>Is the dataset in this competition annotated using the same method as describe in the LiveCell paper? If there is any difference, can we know about that?</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 1571581,
          "author_name": "namgalielei",
          "author_url": "",
          "post_date": "11/05/2021 01:34:54",
          "content": "<p><a href=\"https://www.kaggle.com/christoffersartorius\" target=\"_blank\">@christoffersartorius</a> Can you please answer the above question ? </p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 1570971,
      "author_name": "slawekbiel",
      "author_url": "",
      "post_date": "11/04/2021 14:18:26",
      "content": "<p>FYI, I've gathered things I've found about the dataset here <a href=\"https://www.kaggle.com/c/sartorius-cell-instance-segmentation/discussion/285384\" target=\"_blank\">https://www.kaggle.com/c/sartorius-cell-instance-segmentation/discussion/285384</a></p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 1572159,
      "author_name": "tereka",
      "author_url": "",
      "post_date": "11/05/2021 13:12:29",
      "content": "<p><a href=\"https://www.kaggle.com/christoffersartorius\" target=\"_blank\">@christoffersartorius</a> </p>\n<p>I have a similar question.<br>\nAre LIVECell and this competiton same annotator and same process right?<br>\nIt's important for building processing.</p>",
      "votes": null,
      "replies": [
        {
          "id": 1572222,
          "author_name": "namgalielei",
          "author_url": "",
          "post_date": "11/05/2021 14:25:31",
          "content": "<p>In my observation, for LiveCell dataset, consider only Shsy5y images, there are images with about 1400 instances, which is much higher than that of the competition data. Another point is Shsy5y  LiveCell data has some images looking alike Astro and Cort. </p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 1572350,
          "author_name": "tereka",
          "author_url": "",
          "post_date": "11/05/2021 15:52:24",
          "content": "<p>thanks for your replay.<br>\nshsy5y is maybe good external dataset than other cells.  <br>\nI found same question from you in thread. </p>\n<p>I'm looking forward to answer from host.</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 1572481,
          "author_name": "anlthms",
          "author_url": "",
          "post_date": "11/05/2021 17:47:59",
          "content": "<blockquote>\n  <p>Another point is Shsy5y LiveCell data has some images looking alike Astro and Cort.</p>\n</blockquote>\n<p>Indeed! Here is an example image from the LiveCell dataset. This image (SHSY5Y_Phase_C10_2_00d04h00m_3.tif) is from the livecell_train_val_images/SHSY5Y directory, but does not look anything like SHSY5Y.</p>\n<p><img src=\"https://raw.githubusercontent.com/anlthms/image-repo/main/SHSY5Y_Phase_C10_2_00d04h00m_3.png\" alt=\"\"></p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "1570484": "Does it mean that there is no overlapping images between LIVECell dataset and the competition dataset in term of SH-SY5Y cell line ? @christoffersartorius",
    "1570825": "I am not sure if I get it right.   in some discussion, It states that list of dictionary vs dictionary of dictionary.",
    "1570869": "Hi The Nam.\n\nLIVECell is a publication with we did in nature methods where we presented a resource with 1.6 million annotated cells. From our work with that dataset we noticed that our models struggle with neuronal cells, more specifically the SH-SY5Y.\n\nThe dataset presented in this competition is a product of a second round of gathering data and annotating with the specific purpose of generating stronger models with regards to neuronal cells. The SH-SY5Y images from the two datasets should not overlap.\n\nI hope that answers your questions.\n\nKindly,\nChristoffer",
    "1570935": "Thanks for your reply, I understand now.",
    "1570948": "> LIVECell is a publication with we did in **nature methods**\n\nIs the dataset in this competition annotated using the same method as describe in the LiveCell paper? If there is any difference, can we know about that?",
    "1570971": "FYI, I've gathered things I've found about the dataset here https://www.kaggle.com/c/sartorius-cell-instance-segmentation/discussion/285384",
    "1571581": "christoffersartorius Can you please answer the above question ?",
    "1572159": "christoffersartorius \n\nI have a similar question.\nAre LIVECell and this competiton same annotator and same process right?\nIt's important for building processing.",
    "1572222": "In my observation, for LiveCell dataset, consider only Shsy5y images, there are images with about 1400 instances, which is much higher than that of the competition data. Another point is Shsy5y  LiveCell data has some images looking alike Astro and Cort.",
    "1572350": "thanks for your replay.\nshsy5y is maybe good external dataset than other cells.  \nI found same question from you in thread. \n\nI'm looking forward to answer from host.",
    "1572481": "> Another point is Shsy5y LiveCell data has some images looking alike Astro and Cort.\n\nIndeed! Here is an example image from the LiveCell dataset. This image (SHSY5Y_Phase_C10_2_00d04h00m_3.tif) is from the livecell_train_val_images/SHSY5Y directory, but does not look anything like SHSY5Y.\n\n![](https://raw.githubusercontent.com/anlthms/image-repo/main/SHSY5Y_Phase_C10_2_00d04h00m_3.png)"
  },
  "source": "meta"
}