{
  "id": 115614,
  "title": "Augmentation Techniques: Let's discuss",
  "url": "/competitions/understanding_cloud_organization/discussion/115614",
  "author_name": "",
  "post_date": "2019-11-04T05:03:09.148887600Z",
  "votes": 3,
  "comment_count": 2,
  "views": 0,
  "content": "<p>Currently, I am trying to using augmentation to improve my segmentation model but didn't get many promising results. Here are a few augmentations that I used</p>\n\n<p><code>\n    train_transform = [\n        albu.HorizontalFlip(p=0.5),\n        albu.VerticalFlip(p=0.5)\n        albu.ShiftScaleRotate(scale_limit=0.5, rotate_limit=0, shift_limit=0.1, p=0.5, border_mode=0),\n        albu.Resize(352, 512)\n    ]\n</code>\nI am getting a decent score for every model around 0.644-0.648 but wanted to know what all augmentations can be explored.</p>\n\n<p>What other augmentations worked for others?</p>",
  "messages": [
    {
      "id": "664703",
      "postDate": "11/04/2019 05:03:09",
      "content": "<p>Currently, I am trying to using augmentation to improve my segmentation model but didn't get many promising results. Here are a few augmentations that I used</p>\n\n<p><code>\n    train_transform = [\n        albu.HorizontalFlip(p=0.5),\n        albu.VerticalFlip(p=0.5)\n        albu.ShiftScaleRotate(scale_limit=0.5, rotate_limit=0, shift_limit=0.1, p=0.5, border_mode=0),\n        albu.Resize(352, 512)\n    ]\n</code>\nI am getting a decent score for every model around 0.644-0.648 but wanted to know what all augmentations can be explored.</p>\n\n<p>What other augmentations worked for others?</p>",
      "rawMarkdown": "Currently, I am trying to using augmentation to improve my segmentation model but didn't get many promising results. Here are a few augmentations that I used\n\n```\n    train_transform = [\n        albu.HorizontalFlip(p=0.5),\n        albu.VerticalFlip(p=0.5)\n        albu.ShiftScaleRotate(scale_limit=0.5, rotate_limit=0, shift_limit=0.1, p=0.5, border_mode=0),\n        albu.Resize(352, 512)\n    ]\n```\nI am getting a decent score for every model around 0.644-0.648 but wanted to know what all augmentations can be explored.\n\nWhat other augmentations worked for others?",
      "votes": null
    },
    {
      "id": "664771",
      "postDate": "11/04/2019 07:44:27",
      "content": "<p>Check <a href=\"https://www.kaggle.com/c/understanding_cloud_organization/discussion/109645#latest-664479\">this</a> out</p>",
      "rawMarkdown": "Check [this](https://www.kaggle.com/c/understanding_cloud_organization/discussion/109645#latest-664479) out",
      "votes": null
    },
    {
      "id": "667254",
      "postDate": "11/07/2019 02:45:08",
      "content": "<p>Hello! How is this transform function used？I want to try change input image size during training but I don't konw how to use it</p>",
      "rawMarkdown": "Hello! How is this transform function used？I want to try change input image size during training but I don't konw how to use it",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 664771,
      "author_name": "bibek777",
      "author_url": "",
      "post_date": "11/04/2019 07:44:27",
      "content": "<p>Check <a href=\"https://www.kaggle.com/c/understanding_cloud_organization/discussion/109645#latest-664479\">this</a> out</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 667254,
      "author_name": "gustavo219",
      "author_url": "",
      "post_date": "11/07/2019 02:45:08",
      "content": "<p>Hello! How is this transform function used？I want to try change input image size during training but I don't konw how to use it</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "664703": "Currently, I am trying to using augmentation to improve my segmentation model but didn't get many promising results. Here are a few augmentations that I used\n\n```\n    train_transform = [\n        albu.HorizontalFlip(p=0.5),\n        albu.VerticalFlip(p=0.5)\n        albu.ShiftScaleRotate(scale_limit=0.5, rotate_limit=0, shift_limit=0.1, p=0.5, border_mode=0),\n        albu.Resize(352, 512)\n    ]\n```\nI am getting a decent score for every model around 0.644-0.648 but wanted to know what all augmentations can be explored.\n\nWhat other augmentations worked for others?",
    "664771": "Check [this](https://www.kaggle.com/c/understanding_cloud_organization/discussion/109645#latest-664479) out",
    "667254": "Hello! How is this transform function used？I want to try change input image size during training but I don't konw how to use it"
  },
  "source": "meta"
}