{
  "id": 199025,
  "title": "[Starter Datasets] - True Positive + False Positive Spectrograms.",
  "url": "/competitions/rfcx-species-audio-detection/discussion/199025",
  "author_name": "",
  "post_date": "2020-11-24T04:13:25.605753700Z",
  "votes": 14,
  "comment_count": 1,
  "views": 0,
  "content": "<p>Hi Folks,</p>\n<p>I've created two datasets to help speed up training the networks.</p>\n<p>These datasets were created by looping through the train_tp and train_fp csv files and extracting 10 second spectrograms where the true positive species or false positive species was confirmed.  </p>\n<p><strong>True Positive Spectrograms</strong><br>\n<a href=\"https://www.kaggle.com/tpmeli/true-positive-centered-10-second-spectrograms\" target=\"_blank\">https://www.kaggle.com/tpmeli/true-positive-centered-10-second-spectrograms</a></p>\n<p><strong>False Positive Spectrograms</strong><br>\n<a href=\"https://www.kaggle.com/tpmeli/false-positive-centered-10-second-spectrograms\" target=\"_blank\">https://www.kaggle.com/tpmeli/false-positive-centered-10-second-spectrograms</a></p>\n<p><strong>A Kernel describing how I've extracted these</strong><br>\n<a href=\"https://www.kaggle.com/tpmeli/how-i-extracted-tp-and-fp-spectrograms\" target=\"_blank\">https://www.kaggle.com/tpmeli/how-i-extracted-tp-and-fp-spectrograms</a></p>\n<p>The sound is centered so that there is the same amount of empty space around it so you may need to do horizontal shifts in data augmenting to deal with this.</p>\n<p>Enjoy!</p>",
  "messages": [
    {
      "id": "1088920",
      "postDate": "11/24/2020 04:13:25",
      "content": "<p>Hi Folks,</p>\n<p>I've created two datasets to help speed up training the networks.</p>\n<p>These datasets were created by looping through the train_tp and train_fp csv files and extracting 10 second spectrograms where the true positive species or false positive species was confirmed.  </p>\n<p><strong>True Positive Spectrograms</strong><br>\n<a href=\"https://www.kaggle.com/tpmeli/true-positive-centered-10-second-spectrograms\" target=\"_blank\">https://www.kaggle.com/tpmeli/true-positive-centered-10-second-spectrograms</a></p>\n<p><strong>False Positive Spectrograms</strong><br>\n<a href=\"https://www.kaggle.com/tpmeli/false-positive-centered-10-second-spectrograms\" target=\"_blank\">https://www.kaggle.com/tpmeli/false-positive-centered-10-second-spectrograms</a></p>\n<p><strong>A Kernel describing how I've extracted these</strong><br>\n<a href=\"https://www.kaggle.com/tpmeli/how-i-extracted-tp-and-fp-spectrograms\" target=\"_blank\">https://www.kaggle.com/tpmeli/how-i-extracted-tp-and-fp-spectrograms</a></p>\n<p>The sound is centered so that there is the same amount of empty space around it so you may need to do horizontal shifts in data augmenting to deal with this.</p>\n<p>Enjoy!</p>",
      "rawMarkdown": "Hi Folks,\n\nI've created two datasets to help speed up training the networks.\n\nThese datasets were created by looping through the train_tp and train_fp csv files and extracting 10 second spectrograms where the true positive species or false positive species was confirmed.  \n\n**True Positive Spectrograms**\n[https://www.kaggle.com/tpmeli/true-positive-centered-10-second-spectrograms](https://www.kaggle.com/tpmeli/true-positive-centered-10-second-spectrograms)\n\n**False Positive Spectrograms**\n[https://www.kaggle.com/tpmeli/false-positive-centered-10-second-spectrograms](https://www.kaggle.com/tpmeli/false-positive-centered-10-second-spectrograms)\n\n**A Kernel describing how I've extracted these**\n[https://www.kaggle.com/tpmeli/how-i-extracted-tp-and-fp-spectrograms](https://www.kaggle.com/tpmeli/how-i-extracted-tp-and-fp-spectrograms)\n\n\nThe sound is centered so that there is the same amount of empty space around it so you may need to do horizontal shifts in data augmenting to deal with this.\n\nEnjoy!",
      "votes": null
    },
    {
      "id": "1092023",
      "postDate": "11/26/2020 13:34:53",
      "content": "<p><a href=\"https://www.kaggle.com/tpmeli\" target=\"_blank\">@tpmeli</a>  Wow ..You are sharing helpful resources in all competitions . Thanks for sharing this resource</p>",
      "rawMarkdown": "tpmeli  Wow ..You are sharing helpful resources in all competitions . Thanks for sharing this resource",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1092023,
      "author_name": "usharengaraju",
      "author_url": "",
      "post_date": "11/26/2020 13:34:53",
      "content": "<p><a href=\"https://www.kaggle.com/tpmeli\" target=\"_blank\">@tpmeli</a>  Wow ..You are sharing helpful resources in all competitions . Thanks for sharing this resource</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1088920": "Hi Folks,\n\nI've created two datasets to help speed up training the networks.\n\nThese datasets were created by looping through the train_tp and train_fp csv files and extracting 10 second spectrograms where the true positive species or false positive species was confirmed.  \n\n**True Positive Spectrograms**\n[https://www.kaggle.com/tpmeli/true-positive-centered-10-second-spectrograms](https://www.kaggle.com/tpmeli/true-positive-centered-10-second-spectrograms)\n\n**False Positive Spectrograms**\n[https://www.kaggle.com/tpmeli/false-positive-centered-10-second-spectrograms](https://www.kaggle.com/tpmeli/false-positive-centered-10-second-spectrograms)\n\n**A Kernel describing how I've extracted these**\n[https://www.kaggle.com/tpmeli/how-i-extracted-tp-and-fp-spectrograms](https://www.kaggle.com/tpmeli/how-i-extracted-tp-and-fp-spectrograms)\n\n\nThe sound is centered so that there is the same amount of empty space around it so you may need to do horizontal shifts in data augmenting to deal with this.\n\nEnjoy!",
    "1092023": "tpmeli  Wow ..You are sharing helpful resources in all competitions . Thanks for sharing this resource"
  },
  "source": "meta"
}