{
  "id": 215172,
  "title": "Using HPA Cell Segmentation Model (No Internet)",
  "url": "/competitions/hpa-single-cell-image-classification/discussion/215172",
  "author_name": "RDizzl3",
  "post_date": "2021-01-28T22:54:26.154000",
  "votes": 25,
  "comment_count": 0,
  "views": 0,
  "content": "<p>Hey all,</p>\n<p>One of the repositories <a href=\"https://github.com/CellProfiling/HPA-Cell-Segmentation\" target=\"_blank\">https://github.com/CellProfiling/HPA-Cell-Segmentation</a> released by the hosts of this competition looks pretty awesome so far. When you install the package the issue with it is that there is no internet access (for submissions) and part of the install relies on different downloads. I am making my notebook available that can use this code base off line. There are a number of steps involved in getting this to work offline. I don't know that it would be part of any final solutions but it looks really useful for building cell masks and I am sure can be used in a variety of other ways. I also provide an example on a test set image as well.</p>\n<p>This was my process:</p>\n<ol>\n<li>Gathering the master zip of hpa pytorch zoo</li>\n<li>Modifying the requirements and setup in the hpa cell segmentator. I had to remove the torch installs and instead of using <code>torch==1.4.0</code> I use the default in the kernel <code>1.7.0</code></li>\n<li>Loading the model weights (I provide nuclei and multi channel cell)</li>\n</ol>\n<p>I loaded the above steps as separate datasets. If there is a better way would love to hear but for now check it out and any feedback is appreciated!</p>\n<p>hpa-cell-segmentator-modifed-master: <a href=\"https://www.kaggle.com/rdizzl3/hpacellsegmentatormaster\" target=\"_blank\">https://www.kaggle.com/rdizzl3/hpacellsegmentatormaster</a></p>\n<p>hpa-cell-segmentator-model-weights: <a href=\"https://www.kaggle.com/rdizzl3/hpacellsegmentatormodelweights\" target=\"_blank\">https://www.kaggle.com/rdizzl3/hpacellsegmentatormodelweights</a></p>\n<p>hpa-pytorch-model-zoo:<br>\n<a href=\"https://www.kaggle.com/rdizzl3/hpapytorchzoozip\" target=\"_blank\">https://www.kaggle.com/rdizzl3/hpapytorchzoozip</a></p>\n<p>Notebook to verify and use installs:<br>\n<a href=\"https://www.kaggle.com/rdizzl3/hpa-segmentation-masks-no-internet\" target=\"_blank\">https://www.kaggle.com/rdizzl3/hpa-segmentation-masks-no-internet</a></p>",
  "messages": [
    {
      "id": 1175115,
      "postDate": "2021-01-28T22:54:26.153Z",
      "content": "<p>Hey all,</p>\n<p>One of the repositories <a href=\"https://github.com/CellProfiling/HPA-Cell-Segmentation\" target=\"_blank\">https://github.com/CellProfiling/HPA-Cell-Segmentation</a> released by the hosts of this competition looks pretty awesome so far. When you install the package the issue with it is that there is no internet access (for submissions) and part of the install relies on different downloads. I am making my notebook available that can use this code base off line. There are a number of steps involved in getting this to work offline. I don't know that it would be part of any final solutions but it looks really useful for building cell masks and I am sure can be used in a variety of other ways. I also provide an example on a test set image as well.</p>\n<p>This was my process:</p>\n<ol>\n<li>Gathering the master zip of hpa pytorch zoo</li>\n<li>Modifying the requirements and setup in the hpa cell segmentator. I had to remove the torch installs and instead of using <code>torch==1.4.0</code> I use the default in the kernel <code>1.7.0</code></li>\n<li>Loading the model weights (I provide nuclei and multi channel cell)</li>\n</ol>\n<p>I loaded the above steps as separate datasets. If there is a better way would love to hear but for now check it out and any feedback is appreciated!</p>\n<p>hpa-cell-segmentator-modifed-master: <a href=\"https://www.kaggle.com/rdizzl3/hpacellsegmentatormaster\" target=\"_blank\">https://www.kaggle.com/rdizzl3/hpacellsegmentatormaster</a></p>\n<p>hpa-cell-segmentator-model-weights: <a href=\"https://www.kaggle.com/rdizzl3/hpacellsegmentatormodelweights\" target=\"_blank\">https://www.kaggle.com/rdizzl3/hpacellsegmentatormodelweights</a></p>\n<p>hpa-pytorch-model-zoo:<br>\n<a href=\"https://www.kaggle.com/rdizzl3/hpapytorchzoozip\" target=\"_blank\">https://www.kaggle.com/rdizzl3/hpapytorchzoozip</a></p>\n<p>Notebook to verify and use installs:<br>\n<a href=\"https://www.kaggle.com/rdizzl3/hpa-segmentation-masks-no-internet\" target=\"_blank\">https://www.kaggle.com/rdizzl3/hpa-segmentation-masks-no-internet</a></p>",
      "rawMarkdown": "Hey all,\n\nOne of the repositories https://github.com/CellProfiling/HPA-Cell-Segmentation released by the hosts of this competition looks pretty awesome so far. When you install the package the issue with it is that there is no internet access (for submissions) and part of the install relies on different downloads. I am making my notebook available that can use this code base off line. There are a number of steps involved in getting this to work offline. I don't know that it would be part of any final solutions but it looks really useful for building cell masks and I am sure can be used in a variety of other ways. I also provide an example on a test set image as well.\n\nThis was my process:\n\n1. Gathering the master zip of hpa pytorch zoo\n2. Modifying the requirements and setup in the hpa cell segmentator. I had to remove the torch installs and instead of using `torch==1.4.0` I use the default in the kernel `1.7.0`\n3. Loading the model weights (I provide nuclei and multi channel cell)\n\nI loaded the above steps as separate datasets. If there is a better way would love to hear but for now check it out and any feedback is appreciated!\n\nhpa-cell-segmentator-modifed-master: https://www.kaggle.com/rdizzl3/hpacellsegmentatormaster\n\nhpa-cell-segmentator-model-weights: https://www.kaggle.com/rdizzl3/hpacellsegmentatormodelweights\n\nhpa-pytorch-model-zoo:\nhttps://www.kaggle.com/rdizzl3/hpapytorchzoozip\n\nNotebook to verify and use installs:\nhttps://www.kaggle.com/rdizzl3/hpa-segmentation-masks-no-internet",
      "votes": 25
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "1175115": "Hey all,\n\nOne of the repositories https://github.com/CellProfiling/HPA-Cell-Segmentation released by the hosts of this competition looks pretty awesome so far. When you install the package the issue with it is that there is no internet access (for submissions) and part of the install relies on different downloads. I am making my notebook available that can use this code base off line. There are a number of steps involved in getting this to work offline. I don't know that it would be part of any final solutions but it looks really useful for building cell masks and I am sure can be used in a variety of other ways. I also provide an example on a test set image as well.\n\nThis was my process:\n\n1. Gathering the master zip of hpa pytorch zoo\n2. Modifying the requirements and setup in the hpa cell segmentator. I had to remove the torch installs and instead of using `torch==1.4.0` I use the default in the kernel `1.7.0`\n3. Loading the model weights (I provide nuclei and multi channel cell)\n\nI loaded the above steps as separate datasets. If there is a better way would love to hear but for now check it out and any feedback is appreciated!\n\nhpa-cell-segmentator-modifed-master: https://www.kaggle.com/rdizzl3/hpacellsegmentatormaster\n\nhpa-cell-segmentator-model-weights: https://www.kaggle.com/rdizzl3/hpacellsegmentatormodelweights\n\nhpa-pytorch-model-zoo:\nhttps://www.kaggle.com/rdizzl3/hpapytorchzoozip\n\nNotebook to verify and use installs:\nhttps://www.kaggle.com/rdizzl3/hpa-segmentation-masks-no-internet"
  }
}