{
  "id": 177162,
  "title": "[Update] 3rd place solution",
  "url": "/competitions/landmark-retrieval-2020/discussion/177162",
  "author_name": "Ke",
  "post_date": "2020-08-25T04:24:11.688000",
  "votes": 22,
  "comment_count": 3,
  "views": 0,
  "content": "<p>[Update 2020/10/09] our code: <a href=\"https://github.com/Raykoooo/3rd_place_to_Kaggle_Google_Landmark_Retrieval_2020\" target=\"_blank\">https://github.com/Raykoooo/3rd_place_to_Kaggle_Google_Landmark_Retrieval_2020</a></p>\n<p>[Update] arxiv: <a href=\"https://arxiv.org/abs/2008.10480\" target=\"_blank\">https://arxiv.org/abs/2008.10480</a></p>\n<p>Our solution:<br>\n<a href=\"https://drive.google.com/file/d/1QOeQqVckIlvDAWfAh0E_lrMEqmQR_Bzr/view\" target=\"_blank\">https://drive.google.com/file/d/1QOeQqVckIlvDAWfAh0E_lrMEqmQR_Bzr/view</a></p>\n<p>We have submitted a paper describing our method to arxiv. But it cannot be accessed now. Once the arxiv link of the paper can be accessed, we will immediately post it here. </p>",
  "messages": [
    {
      "id": 984378,
      "postDate": "2020-08-25T04:24:11.690Z",
      "content": "<p>[Update 2020/10/09] our code: <a href=\"https://github.com/Raykoooo/3rd_place_to_Kaggle_Google_Landmark_Retrieval_2020\" target=\"_blank\">https://github.com/Raykoooo/3rd_place_to_Kaggle_Google_Landmark_Retrieval_2020</a></p>\n<p>[Update] arxiv: <a href=\"https://arxiv.org/abs/2008.10480\" target=\"_blank\">https://arxiv.org/abs/2008.10480</a></p>\n<p>Our solution:<br>\n<a href=\"https://drive.google.com/file/d/1QOeQqVckIlvDAWfAh0E_lrMEqmQR_Bzr/view\" target=\"_blank\">https://drive.google.com/file/d/1QOeQqVckIlvDAWfAh0E_lrMEqmQR_Bzr/view</a></p>\n<p>We have submitted a paper describing our method to arxiv. But it cannot be accessed now. Once the arxiv link of the paper can be accessed, we will immediately post it here. </p>",
      "rawMarkdown": "[Update 2020/10/09] our code: https://github.com/Raykoooo/3rd_place_to_Kaggle_Google_Landmark_Retrieval_2020\n\n[Update] arxiv: https://arxiv.org/abs/2008.10480\n\nOur solution:\nhttps://drive.google.com/file/d/1QOeQqVckIlvDAWfAh0E_lrMEqmQR_Bzr/view\n\nWe have submitted a paper describing our method to arxiv. But it cannot be accessed now. Once the arxiv link of the paper can be accessed, we will immediately post it here. \n",
      "votes": 22
    },
    {
      "id": 984434,
      "postDate": "2020-08-25T05:12:34.133Z",
      "content": "<p>Thanks for sharing! Could I please ask - is there a reason why you decided to use corner cutmix instead of cutmix please?</p>",
      "rawMarkdown": "Thanks for sharing! Could I please ask - is there a reason why you decided to use corner cutmix instead of cutmix please?",
      "votes": 1,
      "replies": [
        {
          "id": 984530,
          "postDate": "2020-08-25T06:17:12.983Z",
          "content": "<p>The reason we do this is not to completely destroy the discriminative part of the image.<br>\nFor landmark images, the most discriminative part is usually near the center of the image. Therefore, we only cutmix at the corners of the image to ensure that both image A and image B can be recognized.</p>",
          "rawMarkdown": "The reason we do this is not to completely destroy the discriminative part of the image.\nFor landmark images, the most discriminative part is usually near the center of the image. Therefore, we only cutmix at the corners of the image to ensure that both image A and image B can be recognized.",
          "votes": 2
        }
      ]
    },
    {
      "id": 994954,
      "postDate": "2020-09-02T04:56:16.957Z",
      "rawMarkdown": "",
      "isDeleted": true
    }
  ],
  "comments": [
    {
      "id": 984434,
      "author_name": "Aman Arora",
      "author_url": "",
      "post_date": "2020-08-25T05:12:34.133000",
      "content": "<p>Thanks for sharing! Could I please ask - is there a reason why you decided to use corner cutmix instead of cutmix please?</p>",
      "votes": 1,
      "replies": [
        {
          "id": 984530,
          "author_name": "Ke",
          "author_url": "",
          "post_date": "2020-08-25T06:17:12.983000",
          "content": "<p>The reason we do this is not to completely destroy the discriminative part of the image.<br>\nFor landmark images, the most discriminative part is usually near the center of the image. Therefore, we only cutmix at the corners of the image to ensure that both image A and image B can be recognized.</p>",
          "votes": 2,
          "replies": []
        }
      ]
    },
    {
      "id": 994954,
      "author_name": "",
      "author_url": "",
      "post_date": "2020-09-02T04:56:16.957000",
      "content": "",
      "votes": 0,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "984378": "[Update 2020/10/09] our code: https://github.com/Raykoooo/3rd_place_to_Kaggle_Google_Landmark_Retrieval_2020\n\n[Update] arxiv: https://arxiv.org/abs/2008.10480\n\nOur solution:\nhttps://drive.google.com/file/d/1QOeQqVckIlvDAWfAh0E_lrMEqmQR_Bzr/view\n\nWe have submitted a paper describing our method to arxiv. But it cannot be accessed now. Once the arxiv link of the paper can be accessed, we will immediately post it here. \n",
    "984434": "Thanks for sharing! Could I please ask - is there a reason why you decided to use corner cutmix instead of cutmix please?",
    "994954": ""
  }
}