{
  "id": 414513,
  "title": "how to use mmdet==3.x with python=3.10 in kaggle notebooks?",
  "url": "/competitions/hubmap-hacking-the-human-vasculature/discussion/414513",
  "author_name": "",
  "post_date": "2023-06-02T02:02:40.035235400Z",
  "votes": 6,
  "comment_count": 7,
  "views": 0,
  "content": "<p>I can't install mmcv without internet…</p>\n<pre><code>The detected CUDA version (12.1) mismatches the version that was used to compile\n      PyTorch (11.8). Please make sure to use the same CUDA versions.The detected CUDA version (12.1) mismatches the version that was used to compile\n      PyTorch (11.8). Please make sure to use the same CUDA versions.\n</code></pre>\n<p><a href=\"https://www.kaggle.com/code/abebe9849/mmde-deto/notebookhttps://www.kaggle.com/code/abebe9849/mmde-deto/notebook\" target=\"_blank\">https://www.kaggle.com/code/abebe9849/mmde-deto/notebookhttps://www.kaggle.com/code/abebe9849/mmde-deto/notebook</a> </p>",
  "messages": [
    {
      "id": "2284388",
      "postDate": "06/02/2023 02:02:40",
      "content": "<p>I can't install mmcv without internet…</p>\n<pre><code>The detected CUDA version (12.1) mismatches the version that was used to compile\n      PyTorch (11.8). Please make sure to use the same CUDA versions.The detected CUDA version (12.1) mismatches the version that was used to compile\n      PyTorch (11.8). Please make sure to use the same CUDA versions.\n</code></pre>\n<p><a href=\"https://www.kaggle.com/code/abebe9849/mmde-deto/notebookhttps://www.kaggle.com/code/abebe9849/mmde-deto/notebook\" target=\"_blank\">https://www.kaggle.com/code/abebe9849/mmde-deto/notebookhttps://www.kaggle.com/code/abebe9849/mmde-deto/notebook</a> </p>",
      "rawMarkdown": "I can't install mmcv without internet...\n```\nThe detected CUDA version (12.1) mismatches the version that was used to compile\n      PyTorch (11.8). Please make sure to use the same CUDA versions.The detected CUDA version (12.1) mismatches the version that was used to compile\n      PyTorch (11.8). Please make sure to use the same CUDA versions.\n```\n\nhttps://www.kaggle.com/code/abebe9849/mmde-deto/notebookhttps://www.kaggle.com/code/abebe9849/mmde-deto/notebook",
      "votes": null
    },
    {
      "id": "2284494",
      "postDate": "06/02/2023 05:23:56",
      "content": "<p>same problem, it seems mmcv 2.0.0 is not compatible with CUDA 12.1</p>",
      "rawMarkdown": "same problem, it seems mmcv 2.0.0 is not compatible with CUDA 12.1",
      "votes": null
    },
    {
      "id": "2284841",
      "postDate": "06/02/2023 09:56:38",
      "content": "<p>I don't know about mmcv==2.x but i was able to compile mmcv==1.7.1 using the following commands -</p>\n<pre><code>!git clone -b v1.7.1 https://github.com/open-mmlab/mmcv.git\n!cd mmcv &amp;&amp; CUDA_HOME=/usr/local/cuda-11.8 MMCV_WITH_OPS=1 pip wheel --wheel-dir=/kaggle/working .\n!rm -rf mmcv/ &amp;&amp; python -m pip install mmcv-full --no-index --find-links=/kaggle/working/\n!git clone -b v1.7.1 https://github.com/open-mmlab/mmcv.git\n!cd mmcv &amp;&amp; CUDA_HOME=/usr/local/cuda-11.8 MMCV_WITH_OPS=1 pip wheel --wheel-dir=/kaggle/working .\n!rm -rf mmcv/ &amp;&amp; python -m pip install mmcv-full --no-index --find-links=/kaggle/working/\n</code></pre>\n<p>post which you can use use mmdet 2.x</p>\n<pre><code>!pip install -U mmdet==2.28.2\n</code></pre>",
      "rawMarkdown": "I don't know about mmcv==2.x but i was able to compile mmcv==1.7.1 using the following commands -\n```\n!git clone -b v1.7.1 https://github.com/open-mmlab/mmcv.git\n!cd mmcv && CUDA_HOME=/usr/local/cuda-11.8 MMCV_WITH_OPS=1 pip wheel --wheel-dir=/kaggle/working .\n!rm -rf mmcv/ && python -m pip install mmcv-full --no-index --find-links=/kaggle/working/\n!git clone -b v1.7.1 https://github.com/open-mmlab/mmcv.git\n!cd mmcv && CUDA_HOME=/usr/local/cuda-11.8 MMCV_WITH_OPS=1 pip wheel --wheel-dir=/kaggle/working .\n!rm -rf mmcv/ && python -m pip install mmcv-full --no-index --find-links=/kaggle/working/\n```\npost which you can use use mmdet 2.x\n```\n!pip install -U mmdet==2.28.2\n```",
      "votes": null
    },
    {
      "id": "2285023",
      "postDate": "06/02/2023 11:58:19",
      "content": "<p>Hello. <br>\nIf only problem is offline mode, I found what notebook chaining can solve this problems:</p>\n<p><a href=\"https://www.kaggle.com/tsobolev/prepare-packages-for-offline-usage\" target=\"_blank\">https://www.kaggle.com/tsobolev/prepare-packages-for-offline-usage</a><br>\n<a href=\"https://www.kaggle.com/tsobolev/package-offline-install\" target=\"_blank\">https://www.kaggle.com/tsobolev/package-offline-install</a></p>\n<p><img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F14675197%2Fd8935dfe48b50fd434c87e1b07a432e3%2Fscreen.png?generation=1685711158449472&amp;alt=media\" alt=\"\"></p>",
      "rawMarkdown": "Hello. \nIf only problem is offline mode, I found what notebook chaining can solve this problems:\n\nhttps://www.kaggle.com/tsobolev/prepare-packages-for-offline-usage\nhttps://www.kaggle.com/tsobolev/package-offline-install\n\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F14675197%2Fd8935dfe48b50fd434c87e1b07a432e3%2Fscreen.png?generation=1685711158449472&alt=media)",
      "votes": null
    },
    {
      "id": "2286204",
      "postDate": "06/03/2023 09:49:21",
      "content": "<p>You can build mmcv wheel for python3.10 like this <a href=\"https://www.kaggle.com/code/zzy990106/mmdet3-wheels\" target=\"_blank\">https://www.kaggle.com/code/zzy990106/mmdet3-wheels</a><br>\n<img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F1953180%2F801374d72130879d868fc6cd2ebaa8c9%2F1685785637808.png?generation=1685785720661061&amp;alt=media\" alt=\"\"></p>",
      "rawMarkdown": "You can build mmcv wheel for python3.10 like this https://www.kaggle.com/code/zzy990106/mmdet3-wheels\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F1953180%2F801374d72130879d868fc6cd2ebaa8c9%2F1685785637808.png?generation=1685785720661061&alt=media)",
      "votes": null
    },
    {
      "id": "2302286",
      "postDate": "06/14/2023 12:57:33",
      "content": "<p>I have found that the current environment has PyTorch version 2.0.0 and CUDA version 12.1. I attempted to install mmcv 2.0.0 using the whl file offline, but it failed. I'm wondering if anyone has successfully attempted this installation.</p>",
      "rawMarkdown": "I have found that the current environment has PyTorch version 2.0.0 and CUDA version 12.1. I attempted to install mmcv 2.0.0 using the whl file offline, but it failed. I'm wondering if anyone has successfully attempted this installation.",
      "votes": null
    },
    {
      "id": "2310212",
      "postDate": "06/20/2023 07:54:15",
      "content": "<p>Have you tried mmDet3.x with mmcv2.x and mmseg1？</p>",
      "rawMarkdown": "Have you tried mmDet3.x with mmcv2.x and mmseg1？",
      "votes": null
    },
    {
      "id": "2321181",
      "postDate": "06/28/2023 11:02:13",
      "content": "<p>It seems that there might be an issue with the TTA (Test Time Augmentation) of instance segmentation not being supported in mmdet 3.x.</p>\n<p><a href=\"https://github.com/open-mmlab/mmdetection/tree/main/mmdet/models/test_time_augs/det_tta.py\" target=\"_blank\">https://github.com/open-mmlab/mmdetection/tree/main/mmdet/models/test_time_augs/det_tta.py</a></p>\n<p>**# TODO: support instance segmentation TTA<br>\n        assert data_samples[0].pred_instances.get('masks', None) is None, \\<br>\n            **'<strong>TTA of instance segmentation does not support now</strong>.<strong>'</strong></p>",
      "rawMarkdown": "It seems that there might be an issue with the TTA (Test Time Augmentation) of instance segmentation not being supported in mmdet 3.x.\n\nhttps://github.com/open-mmlab/mmdetection/tree/main/mmdet/models/test_time_augs/det_tta.py\n\n**# TODO: support instance segmentation TTA\n        assert data_samples[0].pred_instances.get('masks', None) is None, \\\n            **'**TTA of instance segmentation does not support now**.**'**",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 2284494,
      "author_name": "ptran1203",
      "author_url": "",
      "post_date": "06/02/2023 05:23:56",
      "content": "<p>same problem, it seems mmcv 2.0.0 is not compatible with CUDA 12.1</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 2284841,
      "author_name": "benihime91",
      "author_url": "",
      "post_date": "06/02/2023 09:56:38",
      "content": "<p>I don't know about mmcv==2.x but i was able to compile mmcv==1.7.1 using the following commands -</p>\n<pre><code>!git clone -b v1.7.1 https://github.com/open-mmlab/mmcv.git\n!cd mmcv &amp;&amp; CUDA_HOME=/usr/local/cuda-11.8 MMCV_WITH_OPS=1 pip wheel --wheel-dir=/kaggle/working .\n!rm -rf mmcv/ &amp;&amp; python -m pip install mmcv-full --no-index --find-links=/kaggle/working/\n!git clone -b v1.7.1 https://github.com/open-mmlab/mmcv.git\n!cd mmcv &amp;&amp; CUDA_HOME=/usr/local/cuda-11.8 MMCV_WITH_OPS=1 pip wheel --wheel-dir=/kaggle/working .\n!rm -rf mmcv/ &amp;&amp; python -m pip install mmcv-full --no-index --find-links=/kaggle/working/\n</code></pre>\n<p>post which you can use use mmdet 2.x</p>\n<pre><code>!pip install -U mmdet==2.28.2\n</code></pre>",
      "votes": null,
      "replies": [
        {
          "id": 2310212,
          "author_name": "chg0901",
          "author_url": "",
          "post_date": "06/20/2023 07:54:15",
          "content": "<p>Have you tried mmDet3.x with mmcv2.x and mmseg1？</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 2285023,
      "author_name": "tsobolev",
      "author_url": "",
      "post_date": "06/02/2023 11:58:19",
      "content": "<p>Hello. <br>\nIf only problem is offline mode, I found what notebook chaining can solve this problems:</p>\n<p><a href=\"https://www.kaggle.com/tsobolev/prepare-packages-for-offline-usage\" target=\"_blank\">https://www.kaggle.com/tsobolev/prepare-packages-for-offline-usage</a><br>\n<a href=\"https://www.kaggle.com/tsobolev/package-offline-install\" target=\"_blank\">https://www.kaggle.com/tsobolev/package-offline-install</a></p>\n<p><img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F14675197%2Fd8935dfe48b50fd434c87e1b07a432e3%2Fscreen.png?generation=1685711158449472&amp;alt=media\" alt=\"\"></p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 2286204,
      "author_name": "zzy990106",
      "author_url": "",
      "post_date": "06/03/2023 09:49:21",
      "content": "<p>You can build mmcv wheel for python3.10 like this <a href=\"https://www.kaggle.com/code/zzy990106/mmdet3-wheels\" target=\"_blank\">https://www.kaggle.com/code/zzy990106/mmdet3-wheels</a><br>\n<img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F1953180%2F801374d72130879d868fc6cd2ebaa8c9%2F1685785637808.png?generation=1685785720661061&amp;alt=media\" alt=\"\"></p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 2302286,
      "author_name": "bent1e",
      "author_url": "",
      "post_date": "06/14/2023 12:57:33",
      "content": "<p>I have found that the current environment has PyTorch version 2.0.0 and CUDA version 12.1. I attempted to install mmcv 2.0.0 using the whl file offline, but it failed. I'm wondering if anyone has successfully attempted this installation.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 2321181,
      "author_name": "atom1231",
      "author_url": "",
      "post_date": "06/28/2023 11:02:13",
      "content": "<p>It seems that there might be an issue with the TTA (Test Time Augmentation) of instance segmentation not being supported in mmdet 3.x.</p>\n<p><a href=\"https://github.com/open-mmlab/mmdetection/tree/main/mmdet/models/test_time_augs/det_tta.py\" target=\"_blank\">https://github.com/open-mmlab/mmdetection/tree/main/mmdet/models/test_time_augs/det_tta.py</a></p>\n<p>**# TODO: support instance segmentation TTA<br>\n        assert data_samples[0].pred_instances.get('masks', None) is None, \\<br>\n            **'<strong>TTA of instance segmentation does not support now</strong>.<strong>'</strong></p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "2284388": "I can't install mmcv without internet...\n```\nThe detected CUDA version (12.1) mismatches the version that was used to compile\n      PyTorch (11.8). Please make sure to use the same CUDA versions.The detected CUDA version (12.1) mismatches the version that was used to compile\n      PyTorch (11.8). Please make sure to use the same CUDA versions.\n```\n\nhttps://www.kaggle.com/code/abebe9849/mmde-deto/notebookhttps://www.kaggle.com/code/abebe9849/mmde-deto/notebook",
    "2284494": "same problem, it seems mmcv 2.0.0 is not compatible with CUDA 12.1",
    "2284841": "I don't know about mmcv==2.x but i was able to compile mmcv==1.7.1 using the following commands -\n```\n!git clone -b v1.7.1 https://github.com/open-mmlab/mmcv.git\n!cd mmcv && CUDA_HOME=/usr/local/cuda-11.8 MMCV_WITH_OPS=1 pip wheel --wheel-dir=/kaggle/working .\n!rm -rf mmcv/ && python -m pip install mmcv-full --no-index --find-links=/kaggle/working/\n!git clone -b v1.7.1 https://github.com/open-mmlab/mmcv.git\n!cd mmcv && CUDA_HOME=/usr/local/cuda-11.8 MMCV_WITH_OPS=1 pip wheel --wheel-dir=/kaggle/working .\n!rm -rf mmcv/ && python -m pip install mmcv-full --no-index --find-links=/kaggle/working/\n```\npost which you can use use mmdet 2.x\n```\n!pip install -U mmdet==2.28.2\n```",
    "2285023": "Hello. \nIf only problem is offline mode, I found what notebook chaining can solve this problems:\n\nhttps://www.kaggle.com/tsobolev/prepare-packages-for-offline-usage\nhttps://www.kaggle.com/tsobolev/package-offline-install\n\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F14675197%2Fd8935dfe48b50fd434c87e1b07a432e3%2Fscreen.png?generation=1685711158449472&alt=media)",
    "2286204": "You can build mmcv wheel for python3.10 like this https://www.kaggle.com/code/zzy990106/mmdet3-wheels\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F1953180%2F801374d72130879d868fc6cd2ebaa8c9%2F1685785637808.png?generation=1685785720661061&alt=media)",
    "2302286": "I have found that the current environment has PyTorch version 2.0.0 and CUDA version 12.1. I attempted to install mmcv 2.0.0 using the whl file offline, but it failed. I'm wondering if anyone has successfully attempted this installation.",
    "2310212": "Have you tried mmDet3.x with mmcv2.x and mmseg1？",
    "2321181": "It seems that there might be an issue with the TTA (Test Time Augmentation) of instance segmentation not being supported in mmdet 3.x.\n\nhttps://github.com/open-mmlab/mmdetection/tree/main/mmdet/models/test_time_augs/det_tta.py\n\n**# TODO: support instance segmentation TTA\n        assert data_samples[0].pred_instances.get('masks', None) is None, \\\n            **'**TTA of instance segmentation does not support now**.**'**"
  },
  "source": "meta"
}