{
  "id": 395225,
  "title": "Tiled Dataset",
  "url": "/competitions/vesuvius-challenge-ink-detection/discussion/395225",
  "author_name": "tmyok",
  "post_date": "2023-03-16T11:45:22.155000",
  "votes": 17,
  "comment_count": 11,
  "views": 0,
  "content": "<p>I created tiled PNG dataset.</p>\n<ul>\n<li>Notebook:&nbsp;<a href=\"https://www.kaggle.com/code/tmyok1984/vcid-tile-png\" target=\"_blank\">https://www.kaggle.com/code/tmyok1984/vcid-tile-png</a></li>\n<li>Dataset:&nbsp;<a href=\"https://www.kaggle.com/datasets/tmyok1984/vcid-tile-images\" target=\"_blank\">https://www.kaggle.com/datasets/tmyok1984/vcid-tile-images</a></li>\n</ul>",
  "messages": [
    {
      "id": 2184447,
      "postDate": "2023-03-16T11:45:22.157Z",
      "content": "<p>I created tiled PNG dataset.</p>\n<ul>\n<li>Notebook:&nbsp;<a href=\"https://www.kaggle.com/code/tmyok1984/vcid-tile-png\" target=\"_blank\">https://www.kaggle.com/code/tmyok1984/vcid-tile-png</a></li>\n<li>Dataset:&nbsp;<a href=\"https://www.kaggle.com/datasets/tmyok1984/vcid-tile-images\" target=\"_blank\">https://www.kaggle.com/datasets/tmyok1984/vcid-tile-images</a></li>\n</ul>",
      "rawMarkdown": "I created tiled PNG dataset.\n\n- Notebook: [https://www.kaggle.com/code/tmyok1984/vcid-tile-png](https://www.kaggle.com/code/tmyok1984/vcid-tile-png)\n- Dataset: [https://www.kaggle.com/datasets/tmyok1984/vcid-tile-images](https://www.kaggle.com/datasets/tmyok1984/vcid-tile-images)",
      "votes": 17
    },
    {
      "id": 2188650,
      "postDate": "2023-03-19T20:59:27.167Z",
      "content": "<p>Awesome work, thank you!</p>\n<p>Can you please explain how you obtained files which memory is about 5GB? That's where .png conversion comes in [I think].<br>\nI am also trying to understand whether we lose any information or not after this conversion.</p>",
      "rawMarkdown": "Awesome work, thank you!\n\nCan you please explain how you obtained files which memory is about 5GB? That's where .png conversion comes in [I think].\nI am also trying to understand whether we lose any information or not after this conversion.",
      "votes": 1,
      "replies": [
        {
          "id": 2189006,
          "postDate": "2023-03-20T06:52:40.703Z",
          "content": "<p>The entire process of creating that dataset is described in the notebook below. If you have any questions, please do not hesitate to ask.<br>\n<a href=\"https://www.kaggle.com/code/tmyok1984/vcid-tile-png\" target=\"_blank\">https://www.kaggle.com/code/tmyok1984/vcid-tile-png</a></p>",
          "rawMarkdown": "The entire process of creating that dataset is described in the notebook below. If you have any questions, please do not hesitate to ask.\nhttps://www.kaggle.com/code/tmyok1984/vcid-tile-png",
          "votes": 2,
          "replies": [
            {
              "id": 2189611,
              "postDate": "2023-03-20T16:26:14.420Z",
              "content": "<p>I screened your code and it seems that you are not doing any operation which reduces memory of files. I think that this png conversion can have significant influence on photo quality. If anybody knows if there is the difference between tiff and png files please add a short answer under this thread</p>",
              "rawMarkdown": "I screened your code and it seems that you are not doing any operation which reduces memory of files. I think that this png conversion can have significant influence on photo quality. If anybody knows if there is the difference between tiff and png files please add a short answer under this thread",
              "votes": 2
            },
            {
              "id": 2189901,
              "postDate": "2023-03-20T21:55:21.170Z",
              "rawMarkdown": "",
              "votes": 1,
              "isDeleted": true
            },
            {
              "id": 2189903,
              "postDate": "2023-03-20T21:56:39.960Z",
              "rawMarkdown": "",
              "isDeleted": true
            },
            {
              "id": 2189905,
              "postDate": "2023-03-20T21:57:40.693Z",
              "content": "<p>The original tiff image is 16-bit grayscale. I converted it to 8-bit grayscale when creating the PNG file.</p>\n<pre><code>cv2.imwrite(str(output_image_dir / f\"{i:02d}.png\"), (image_roi * 255).astype(np.uint8))\n</code></pre>",
              "rawMarkdown": "The original tiff image is 16-bit grayscale. I converted it to 8-bit grayscale when creating the PNG file.\n\n```\ncv2.imwrite(str(output_image_dir / f\"{i:02d}.png\"), (image_roi * 255).astype(np.uint8))\n```",
              "votes": 3
            }
          ]
        }
      ]
    },
    {
      "id": 2185130,
      "postDate": "2023-03-16T20:41:52.903Z",
      "content": "<p>Thanks for this! <br>\nHow can we match tiles with the given masks? <br>\nI think we're missing a metadata .csv with tiles x,y and 'Predicted'</p>",
      "rawMarkdown": "Thanks for this! \nHow can we match tiles with the given masks? \nI think we're missing a metadata .csv with tiles x,y and 'Predicted'",
      "votes": 1,
      "replies": [
        {
          "id": 2185182,
          "postDate": "2023-03-16T21:31:59.910Z",
          "content": "<p>Sorry, I forgot to prepare the mask. It might take a while, can you please wait?</p>",
          "rawMarkdown": "Sorry, I forgot to prepare the mask. It might take a while, can you please wait?",
          "votes": 2,
          "replies": [
            {
              "id": 2186514,
              "postDate": "2023-03-17T20:49:55.353Z",
              "content": "<p>I added the mask and train.csv files. Can you please confirm that they're there?</p>",
              "rawMarkdown": "I added the mask and train.csv files. Can you please confirm that they're there?",
              "votes": 1
            }
          ]
        }
      ]
    },
    {
      "id": 2184762,
      "postDate": "2023-03-16T15:50:09.023Z",
      "content": "<p>Awesome, that can come in handy, thank you!</p>",
      "rawMarkdown": "Awesome, that can come in handy, thank you!",
      "votes": 1
    },
    {
      "id": 2184819,
      "postDate": "2023-03-16T16:16:25.643Z",
      "rawMarkdown": "",
      "votes": 1,
      "isDeleted": true
    }
  ],
  "comments": [
    {
      "id": 2188650,
      "author_name": "Maciej Król",
      "author_url": "",
      "post_date": "2023-03-19T20:59:27.167000",
      "content": "<p>Awesome work, thank you!</p>\n<p>Can you please explain how you obtained files which memory is about 5GB? That's where .png conversion comes in [I think].<br>\nI am also trying to understand whether we lose any information or not after this conversion.</p>",
      "votes": 1,
      "replies": [
        {
          "id": 2189006,
          "author_name": "tmyok",
          "author_url": "",
          "post_date": "2023-03-20T06:52:40.703000",
          "content": "<p>The entire process of creating that dataset is described in the notebook below. If you have any questions, please do not hesitate to ask.<br>\n<a href=\"https://www.kaggle.com/code/tmyok1984/vcid-tile-png\" target=\"_blank\">https://www.kaggle.com/code/tmyok1984/vcid-tile-png</a></p>",
          "votes": 2,
          "replies": [
            {
              "id": 2189611,
              "author_name": "Maciej Król",
              "author_url": "",
              "post_date": "2023-03-20T16:26:14.420000",
              "content": "<p>I screened your code and it seems that you are not doing any operation which reduces memory of files. I think that this png conversion can have significant influence on photo quality. If anybody knows if there is the difference between tiff and png files please add a short answer under this thread</p>",
              "votes": 2,
              "replies": []
            },
            {
              "id": 2189901,
              "author_name": "",
              "author_url": "",
              "post_date": "2023-03-20T21:55:21.170000",
              "content": "",
              "votes": 1,
              "replies": []
            },
            {
              "id": 2189903,
              "author_name": "",
              "author_url": "",
              "post_date": "2023-03-20T21:56:39.960000",
              "content": "",
              "votes": 0,
              "replies": []
            },
            {
              "id": 2189905,
              "author_name": "tmyok",
              "author_url": "",
              "post_date": "2023-03-20T21:57:40.693000",
              "content": "<p>The original tiff image is 16-bit grayscale. I converted it to 8-bit grayscale when creating the PNG file.</p>\n<pre><code>cv2.imwrite(str(output_image_dir / f\"{i:02d}.png\"), (image_roi * 255).astype(np.uint8))\n</code></pre>",
              "votes": 3,
              "replies": []
            }
          ]
        }
      ]
    },
    {
      "id": 2185130,
      "author_name": "Ioannis M",
      "author_url": "",
      "post_date": "2023-03-16T20:41:52.903000",
      "content": "<p>Thanks for this! <br>\nHow can we match tiles with the given masks? <br>\nI think we're missing a metadata .csv with tiles x,y and 'Predicted'</p>",
      "votes": 1,
      "replies": [
        {
          "id": 2185182,
          "author_name": "tmyok",
          "author_url": "",
          "post_date": "2023-03-16T21:31:59.910000",
          "content": "<p>Sorry, I forgot to prepare the mask. It might take a while, can you please wait?</p>",
          "votes": 2,
          "replies": [
            {
              "id": 2186514,
              "author_name": "tmyok",
              "author_url": "",
              "post_date": "2023-03-17T20:49:55.353000",
              "content": "<p>I added the mask and train.csv files. Can you please confirm that they're there?</p>",
              "votes": 1,
              "replies": []
            }
          ]
        }
      ]
    },
    {
      "id": 2184762,
      "author_name": "JP Posma",
      "author_url": "",
      "post_date": "2023-03-16T15:50:09.023000",
      "content": "<p>Awesome, that can come in handy, thank you!</p>",
      "votes": 1,
      "replies": []
    },
    {
      "id": 2184819,
      "author_name": "",
      "author_url": "",
      "post_date": "2023-03-16T16:16:25.643000",
      "content": "",
      "votes": 1,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "2184447": "I created tiled PNG dataset.\n\n- Notebook: [https://www.kaggle.com/code/tmyok1984/vcid-tile-png](https://www.kaggle.com/code/tmyok1984/vcid-tile-png)\n- Dataset: [https://www.kaggle.com/datasets/tmyok1984/vcid-tile-images](https://www.kaggle.com/datasets/tmyok1984/vcid-tile-images)",
    "2188650": "Awesome work, thank you!\n\nCan you please explain how you obtained files which memory is about 5GB? That's where .png conversion comes in [I think].\nI am also trying to understand whether we lose any information or not after this conversion.",
    "2185130": "Thanks for this! \nHow can we match tiles with the given masks? \nI think we're missing a metadata .csv with tiles x,y and 'Predicted'",
    "2184762": "Awesome, that can come in handy, thank you!",
    "2184819": ""
  }
}