{
  "id": 163481,
  "title": "GPU not being used",
  "url": "/competitions/siim-isic-melanoma-classification/discussion/163481",
  "author_name": "",
  "post_date": "2020-07-02T08:14:37.219093500Z",
  "votes": 4,
  "comment_count": 7,
  "views": 0,
  "content": "<p><img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F4241581%2Fdbdbf7868264bbaba62d0673e8571887%2FAnnotation%202020-07-02%20134147.jpg?generation=1593677716482783&amp;alt=media\" alt=\"\">\nHello, I am new here. I want to know how to make use of the GPU. Here's a sceenshot of my work.</p>",
  "messages": [
    {
      "id": "912096",
      "postDate": "07/02/2020 08:14:37",
      "content": "<p><img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F4241581%2Fdbdbf7868264bbaba62d0673e8571887%2FAnnotation%202020-07-02%20134147.jpg?generation=1593677716482783&amp;alt=media\" alt=\"\">\nHello, I am new here. I want to know how to make use of the GPU. Here's a sceenshot of my work.</p>",
      "rawMarkdown": "![](https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F4241581%2Fdbdbf7868264bbaba62d0673e8571887%2FAnnotation%202020-07-02%20134147.jpg?generation=1593677716482783&amp;alt=media)\nHello, I am new here. I want to know how to make use of the GPU. Here's a sceenshot of my work.",
      "votes": null
    },
    {
      "id": "912117",
      "postDate": "07/02/2020 08:35:23",
      "content": "<p>It is being used but your code spends far more time just getting and resizing the images. The best way to speed it up is to preprocess the images to the right size first and save them as a dataset or use one of the datasets that other Kagglers have made already such as <a href=\"https://www.kaggle.com/arroqc/siic-isic-224x224-images\">these</a>.</p>",
      "rawMarkdown": "It is being used but your code spends far more time just getting and resizing the images. The best way to speed it up is to preprocess the images to the right size first and save them as a dataset or use one of the datasets that other Kagglers have made already such as [these](https://www.kaggle.com/arroqc/siic-isic-224x224-images).",
      "votes": null
    },
    {
      "id": "912189",
      "postDate": "07/02/2020 09:38:53",
      "content": "<p>Yep, as Bruce said, if you use images directly from the input, it will take a lot of time in loading and resizing, just to add to the list, I had shared some resized images also for 368, 512, 768 and 1024 <a href=\"https://www.kaggle.com/c/siim-isic-melanoma-classification/discussion/161043\">here</a>.</p>\n\n<p>Hope you find it useful! :)</p>",
      "rawMarkdown": "Yep, as Bruce said, if you use images directly from the input, it will take a lot of time in loading and resizing, just to add to the list, I had shared some resized images also for 368, 512, 768 and 1024 [here](https://www.kaggle.com/c/siim-isic-melanoma-classification/discussion/161043).\n\nHope you find it useful! :)",
      "votes": null
    },
    {
      "id": "912542",
      "postDate": "07/02/2020 14:58:46",
      "content": "<p>Thanks mate!\nMy problem has been resolved with this :)</p>",
      "rawMarkdown": "Thanks mate!\nMy problem has been resolved with this :)",
      "votes": null
    },
    {
      "id": "912543",
      "postDate": "07/02/2020 14:58:56",
      "content": "<p>Thanks mate!\nMy problem has been resolved with this :)</p>",
      "rawMarkdown": "Thanks mate!\nMy problem has been resolved with this :)",
      "votes": null
    },
    {
      "id": "912635",
      "postDate": "07/02/2020 16:07:16",
      "content": "<p>Cool! :)</p>",
      "rawMarkdown": "Cool! :)",
      "votes": null
    },
    {
      "id": "1245237",
      "postDate": "03/19/2021 15:36:17",
      "content": "<p><a href=\"https://www.kaggle.com/mutantspore\" target=\"_blank\">@mutantspore</a> I am facing the same problem but I am working on a different dataset. Could you please let me know how I can save the pre-processed images? I am using PyTorch and DataLoader.</p>",
      "rawMarkdown": "mutantspore I am facing the same problem but I am working on a different dataset. Could you please let me know how I can save the pre-processed images? I am using PyTorch and DataLoader.",
      "votes": null
    },
    {
      "id": "2223515",
      "postDate": "04/16/2023 10:47:00",
      "content": "<p>me too. same problem with dataloader</p>",
      "rawMarkdown": "me too. same problem with dataloader",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 912117,
      "author_name": "mutantspore",
      "author_url": "",
      "post_date": "07/02/2020 08:35:23",
      "content": "<p>It is being used but your code spends far more time just getting and resizing the images. The best way to speed it up is to preprocess the images to the right size first and save them as a dataset or use one of the datasets that other Kagglers have made already such as <a href=\"https://www.kaggle.com/arroqc/siic-isic-224x224-images\">these</a>.</p>",
      "votes": null,
      "replies": [
        {
          "id": 912543,
          "author_name": "rohitagarwal08",
          "author_url": "",
          "post_date": "07/02/2020 14:58:56",
          "content": "<p>Thanks mate!\nMy problem has been resolved with this :)</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 1245237,
          "author_name": "shambhavimalik",
          "author_url": "",
          "post_date": "03/19/2021 15:36:17",
          "content": "<p><a href=\"https://www.kaggle.com/mutantspore\" target=\"_blank\">@mutantspore</a> I am facing the same problem but I am working on a different dataset. Could you please let me know how I can save the pre-processed images? I am using PyTorch and DataLoader.</p>",
          "votes": null,
          "replies": [
            {
              "id": 2223515,
              "author_name": "alelat",
              "author_url": "",
              "post_date": "04/16/2023 10:47:00",
              "content": "<p>me too. same problem with dataloader</p>",
              "votes": null,
              "replies": []
            }
          ]
        }
      ]
    },
    {
      "id": 912189,
      "author_name": "sarques",
      "author_url": "",
      "post_date": "07/02/2020 09:38:53",
      "content": "<p>Yep, as Bruce said, if you use images directly from the input, it will take a lot of time in loading and resizing, just to add to the list, I had shared some resized images also for 368, 512, 768 and 1024 <a href=\"https://www.kaggle.com/c/siim-isic-melanoma-classification/discussion/161043\">here</a>.</p>\n\n<p>Hope you find it useful! :)</p>",
      "votes": null,
      "replies": [
        {
          "id": 912542,
          "author_name": "rohitagarwal08",
          "author_url": "",
          "post_date": "07/02/2020 14:58:46",
          "content": "<p>Thanks mate!\nMy problem has been resolved with this :)</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 912635,
          "author_name": "sarques",
          "author_url": "",
          "post_date": "07/02/2020 16:07:16",
          "content": "<p>Cool! :)</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "912096": "![](https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F4241581%2Fdbdbf7868264bbaba62d0673e8571887%2FAnnotation%202020-07-02%20134147.jpg?generation=1593677716482783&amp;alt=media)\nHello, I am new here. I want to know how to make use of the GPU. Here's a sceenshot of my work.",
    "912117": "It is being used but your code spends far more time just getting and resizing the images. The best way to speed it up is to preprocess the images to the right size first and save them as a dataset or use one of the datasets that other Kagglers have made already such as [these](https://www.kaggle.com/arroqc/siic-isic-224x224-images).",
    "912189": "Yep, as Bruce said, if you use images directly from the input, it will take a lot of time in loading and resizing, just to add to the list, I had shared some resized images also for 368, 512, 768 and 1024 [here](https://www.kaggle.com/c/siim-isic-melanoma-classification/discussion/161043).\n\nHope you find it useful! :)",
    "912542": "Thanks mate!\nMy problem has been resolved with this :)",
    "912543": "Thanks mate!\nMy problem has been resolved with this :)",
    "912635": "Cool! :)",
    "1245237": "mutantspore I am facing the same problem but I am working on a different dataset. Could you please let me know how I can save the pre-processed images? I am using PyTorch and DataLoader.",
    "2223515": "me too. same problem with dataloader"
  },
  "source": "meta"
}