{
  "id": 287638,
  "title": "Why Unet is suitable for this competition?",
  "url": "/competitions/sartorius-cell-instance-segmentation/discussion/287638",
  "author_name": "",
  "post_date": "2021-11-15T01:41:28.593361Z",
  "votes": 2,
  "comment_count": 2,
  "views": 0,
  "content": "<p>I see many NB using Unet for instance segmentation, but I am confused how to divide a mask into pieces for every instance. anyone can help explain?</p>",
  "messages": [
    {
      "id": "1582524",
      "postDate": "11/15/2021 01:41:28",
      "content": "<p>I see many NB using Unet for instance segmentation, but I am confused how to divide a mask into pieces for every instance. anyone can help explain?</p>",
      "rawMarkdown": "I see many NB using Unet for instance segmentation, but I am confused how to divide a mask into pieces for every instance. anyone can help explain?",
      "votes": null
    },
    {
      "id": "1582620",
      "postDate": "11/15/2021 05:07:35",
      "content": "<p>Mainly opencv <code>connectedComponents</code> is used to separate the segmented cells from an image. You can see the post-processing part of this NB <a href=\"https://www.kaggle.com/julian3833/sartorius-starter-baseline-torch-u-net#Predict\" target=\"_blank\">🦠 Sartorius - Starter Baseline Torch U-net</a>  by <a href=\"https://www.kaggle.com/julian3833\" target=\"_blank\">@julian3833</a> in the Predict/Utilities part. I also took that code for preprocessing for my NB.</p>\n<p>Check out more about Unet utilization, here</p>\n<ul>\n<li><a href=\"https://www.kaggle.com/c/sartorius-cell-instance-segmentation/discussion/286553\" target=\"_blank\">UNet Strikes Back 🔥 | LB: 0.155+</a></li>\n<li><a href=\"https://www.kaggle.com/c/sartorius-cell-instance-segmentation/discussion/279790\" target=\"_blank\">Tips in submission and baseline (in the beginning of the competition)</a></li>\n<li><a href=\"https://www.kaggle.com/c/sartorius-cell-instance-segmentation/discussion/285422\" target=\"_blank\">Train Attention based Residual Unet with watershed algorithm for instance segmentation</a> [mainly check out the comments ]</li>\n</ul>",
      "rawMarkdown": "Mainly opencv `connectedComponents` is used to separate the segmented cells from an image. You can see the post-processing part of this NB [🦠 Sartorius - Starter Baseline Torch U-net](https://www.kaggle.com/julian3833/sartorius-starter-baseline-torch-u-net#Predict)  by @julian3833 in the Predict/Utilities part. I also took that code for preprocessing for my NB.\n\nCheck out more about Unet utilization, here\n- [UNet Strikes Back 🔥 | LB: 0.155+](https://www.kaggle.com/c/sartorius-cell-instance-segmentation/discussion/286553)\n- [Tips in submission and baseline (in the beginning of the competition)](https://www.kaggle.com/c/sartorius-cell-instance-segmentation/discussion/279790)\n- [Train Attention based Residual Unet with watershed algorithm for instance segmentation](https://www.kaggle.com/c/sartorius-cell-instance-segmentation/discussion/285422) [mainly check out the comments ]",
      "votes": null
    },
    {
      "id": "1582819",
      "postDate": "11/15/2021 10:01:52",
      "content": "<p><a href=\"https://www.kaggle.com/soumya9977\" target=\"_blank\">@soumya9977</a> many thanks👍, these notebooks are super cool!</p>",
      "rawMarkdown": "soumya9977 many thanks👍, these notebooks are super cool!",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1582620,
      "author_name": "soumya9977",
      "author_url": "",
      "post_date": "11/15/2021 05:07:35",
      "content": "<p>Mainly opencv <code>connectedComponents</code> is used to separate the segmented cells from an image. You can see the post-processing part of this NB <a href=\"https://www.kaggle.com/julian3833/sartorius-starter-baseline-torch-u-net#Predict\" target=\"_blank\">🦠 Sartorius - Starter Baseline Torch U-net</a>  by <a href=\"https://www.kaggle.com/julian3833\" target=\"_blank\">@julian3833</a> in the Predict/Utilities part. I also took that code for preprocessing for my NB.</p>\n<p>Check out more about Unet utilization, here</p>\n<ul>\n<li><a href=\"https://www.kaggle.com/c/sartorius-cell-instance-segmentation/discussion/286553\" target=\"_blank\">UNet Strikes Back 🔥 | LB: 0.155+</a></li>\n<li><a href=\"https://www.kaggle.com/c/sartorius-cell-instance-segmentation/discussion/279790\" target=\"_blank\">Tips in submission and baseline (in the beginning of the competition)</a></li>\n<li><a href=\"https://www.kaggle.com/c/sartorius-cell-instance-segmentation/discussion/285422\" target=\"_blank\">Train Attention based Residual Unet with watershed algorithm for instance segmentation</a> [mainly check out the comments ]</li>\n</ul>",
      "votes": null,
      "replies": [
        {
          "id": 1582819,
          "author_name": "plugin1689",
          "author_url": "",
          "post_date": "11/15/2021 10:01:52",
          "content": "<p><a href=\"https://www.kaggle.com/soumya9977\" target=\"_blank\">@soumya9977</a> many thanks👍, these notebooks are super cool!</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "1582524": "I see many NB using Unet for instance segmentation, but I am confused how to divide a mask into pieces for every instance. anyone can help explain?",
    "1582620": "Mainly opencv `connectedComponents` is used to separate the segmented cells from an image. You can see the post-processing part of this NB [🦠 Sartorius - Starter Baseline Torch U-net](https://www.kaggle.com/julian3833/sartorius-starter-baseline-torch-u-net#Predict)  by @julian3833 in the Predict/Utilities part. I also took that code for preprocessing for my NB.\n\nCheck out more about Unet utilization, here\n- [UNet Strikes Back 🔥 | LB: 0.155+](https://www.kaggle.com/c/sartorius-cell-instance-segmentation/discussion/286553)\n- [Tips in submission and baseline (in the beginning of the competition)](https://www.kaggle.com/c/sartorius-cell-instance-segmentation/discussion/279790)\n- [Train Attention based Residual Unet with watershed algorithm for instance segmentation](https://www.kaggle.com/c/sartorius-cell-instance-segmentation/discussion/285422) [mainly check out the comments ]",
    "1582819": "soumya9977 many thanks👍, these notebooks are super cool!"
  },
  "source": "meta"
}