{
  "id": 426393,
  "title": "Releasing unlabeled data of Bangla Documents and their yolo inferred bounding box and class labels.",
  "url": "/competitions/dlsprint2/discussion/426393",
  "author_name": "Sameen53",
  "post_date": "2023-07-23T08:09:57.459000",
  "votes": 1,
  "comment_count": 0,
  "views": 0,
  "content": "<p><a href=\"https://www.kaggle.com/datasets/intesurahmed/badlad-inference-pseudolabels-part-7\" target=\"_blank\">https://www.kaggle.com/datasets/intesurahmed/badlad-inference-pseudolabels-part-7</a></p>\n<h3>About the data</h3>\n<p>The data has been split into 7 parts due to the 100GB limit of Kaggle datasets. The first 6 dataset contains the images of Bengali Documents and the 7th dataset contains the labels generated by yolo.</p>\n<h3>Credit</h3>\n<p>Our wonderful co-hosts Bengali.AI 🥰</p>\n<h3>Caution</h3>\n<p>The labels contain only the class label and the bounding box and not the fine-grained mask. These are machine outputs and obviously not equivalent to human annotations on the main Competition datasets. It remains upto the p the participants to leverage the data if they choose to. Some possible methods were hinted at in the last workshop by Bengali.AI. Check the Facebook page for the recording.</p>\n<p><strong>It's not mandatory to utilize this data.</strong></p>",
  "messages": [
    {
      "id": 2355254,
      "postDate": "2023-07-23T08:09:57.460Z",
      "content": "<p><a href=\"https://www.kaggle.com/datasets/intesurahmed/badlad-inference-pseudolabels-part-7\" target=\"_blank\">https://www.kaggle.com/datasets/intesurahmed/badlad-inference-pseudolabels-part-7</a></p>\n<h3>About the data</h3>\n<p>The data has been split into 7 parts due to the 100GB limit of Kaggle datasets. The first 6 dataset contains the images of Bengali Documents and the 7th dataset contains the labels generated by yolo.</p>\n<h3>Credit</h3>\n<p>Our wonderful co-hosts Bengali.AI 🥰</p>\n<h3>Caution</h3>\n<p>The labels contain only the class label and the bounding box and not the fine-grained mask. These are machine outputs and obviously not equivalent to human annotations on the main Competition datasets. It remains upto the p the participants to leverage the data if they choose to. Some possible methods were hinted at in the last workshop by Bengali.AI. Check the Facebook page for the recording.</p>\n<p><strong>It's not mandatory to utilize this data.</strong></p>",
      "rawMarkdown": "https://www.kaggle.com/datasets/intesurahmed/badlad-inference-pseudolabels-part-7\n\n### About the data\nThe data has been split into 7 parts due to the 100GB limit of Kaggle datasets. The first 6 dataset contains the images of Bengali Documents and the 7th dataset contains the labels generated by yolo.\n\n### Credit\nOur wonderful co-hosts Bengali.AI 🥰\n\n### Caution\nThe labels contain only the class label and the bounding box and not the fine-grained mask. These are machine outputs and obviously not equivalent to human annotations on the main Competition datasets. It remains upto the p the participants to leverage the data if they choose to. Some possible methods were hinted at in the last workshop by Bengali.AI. Check the Facebook page for the recording.\n\n**It's not mandatory to utilize this data.**",
      "votes": 1
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "2355254": "https://www.kaggle.com/datasets/intesurahmed/badlad-inference-pseudolabels-part-7\n\n### About the data\nThe data has been split into 7 parts due to the 100GB limit of Kaggle datasets. The first 6 dataset contains the images of Bengali Documents and the 7th dataset contains the labels generated by yolo.\n\n### Credit\nOur wonderful co-hosts Bengali.AI 🥰\n\n### Caution\nThe labels contain only the class label and the bounding box and not the fine-grained mask. These are machine outputs and obviously not equivalent to human annotations on the main Competition datasets. It remains upto the p the participants to leverage the data if they choose to. Some possible methods were hinted at in the last workshop by Bengali.AI. Check the Facebook page for the recording.\n\n**It's not mandatory to utilize this data.**"
  }
}