{
  "id": 95236,
  "title": "Post-competition Mask-RCNN discussion.",
  "url": "/competitions/imaterialist-fashion-2019-FGVC6/discussion/95236",
  "author_name": "Anish Agnihotri",
  "post_date": "2019-06-11T00:27:01.862000",
  "votes": 0,
  "comment_count": 0,
  "views": 0,
  "content": "<p>I deviated from using vanilla Mask-RCNN near the end of the competition, but it seems that a large majority of the bronze-medalist competitors made their own modifications to the kernel which was shared early in the competition. </p>\n\n<p>I'm curious in knowing what your learning rate/steps and additional additions/augmentations were, and what resulting private LB you got from them.</p>\n\n<p>I'm learning as I go, and just looking to get a general insight.</p>\n\n<p>Cheers, and congratulations to all the fantastic winners!</p>\n\n<p>Anish</p>",
  "messages": [
    {
      "id": 549707,
      "postDate": "2019-06-11T00:27:01.863Z",
      "content": "<p>I deviated from using vanilla Mask-RCNN near the end of the competition, but it seems that a large majority of the bronze-medalist competitors made their own modifications to the kernel which was shared early in the competition. </p>\n\n<p>I'm curious in knowing what your learning rate/steps and additional additions/augmentations were, and what resulting private LB you got from them.</p>\n\n<p>I'm learning as I go, and just looking to get a general insight.</p>\n\n<p>Cheers, and congratulations to all the fantastic winners!</p>\n\n<p>Anish</p>",
      "rawMarkdown": "I deviated from using vanilla Mask-RCNN near the end of the competition, but it seems that a large majority of the bronze-medalist competitors made their own modifications to the kernel which was shared early in the competition. \n\nI'm curious in knowing what your learning rate/steps and additional additions/augmentations were, and what resulting private LB you got from them.\n\nI'm learning as I go, and just looking to get a general insight.\n\nCheers, and congratulations to all the fantastic winners!\n\nAnish"
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "549707": "I deviated from using vanilla Mask-RCNN near the end of the competition, but it seems that a large majority of the bronze-medalist competitors made their own modifications to the kernel which was shared early in the competition. \n\nI'm curious in knowing what your learning rate/steps and additional additions/augmentations were, and what resulting private LB you got from them.\n\nI'm learning as I go, and just looking to get a general insight.\n\nCheers, and congratulations to all the fantastic winners!\n\nAnish"
  }
}