{
  "id": 115086,
  "title": "Code to ensemble boxes for object detection",
  "url": "/competitions/open-images-2019-object-detection/discussion/115086",
  "author_name": "ZFTurbo",
  "post_date": "2019-10-31T06:43:43.981000",
  "votes": 34,
  "comment_count": 5,
  "views": 0,
  "content": "<p>I made a release of Python module for ensembling boxes for Object Detection models. It includes very promicing Weighted Boxes Fusion (WBF) method, which we used instead of NMS in 2019 and 2018 competitions.</p>\n\n<p><a href=\"https://github.com/ZFTurbo/Weighted-Boxes-Fusion\">https://github.com/ZFTurbo/Weighted-Boxes-Fusion</a></p>\n\n<p>Repository contains the following methods:\n- Non-maximum Suppression (NMS)\n- Soft-NMS\n- Non-maximum weighted (NMW)\n- Weighted boxes fusion (WBF)</p>\n\n<p>Installation available using pip: <code>pip install ensemble-boxes</code></p>\n\n<p>It will be interesting to hear if WBF works better for other teams too. )</p>",
  "messages": [
    {
      "id": 662160,
      "postDate": "2019-10-31T06:43:43.980Z",
      "content": "<p>I made a release of Python module for ensembling boxes for Object Detection models. It includes very promicing Weighted Boxes Fusion (WBF) method, which we used instead of NMS in 2019 and 2018 competitions.</p>\n\n<p><a href=\"https://github.com/ZFTurbo/Weighted-Boxes-Fusion\">https://github.com/ZFTurbo/Weighted-Boxes-Fusion</a></p>\n\n<p>Repository contains the following methods:\n- Non-maximum Suppression (NMS)\n- Soft-NMS\n- Non-maximum weighted (NMW)\n- Weighted boxes fusion (WBF)</p>\n\n<p>Installation available using pip: <code>pip install ensemble-boxes</code></p>\n\n<p>It will be interesting to hear if WBF works better for other teams too. )</p>",
      "rawMarkdown": "I made a release of Python module for ensembling boxes for Object Detection models. It includes very promicing Weighted Boxes Fusion (WBF) method, which we used instead of NMS in 2019 and 2018 competitions.\n\nhttps://github.com/ZFTurbo/Weighted-Boxes-Fusion\n\nRepository contains the following methods:\n- Non-maximum Suppression (NMS)\n- Soft-NMS\n- Non-maximum weighted (NMW)\n- Weighted boxes fusion (WBF)\n\nInstallation available using pip: `pip install ensemble-boxes`\n\nIt will be interesting to hear if WBF works better for other teams too. )",
      "votes": 34
    },
    {
      "id": 758617,
      "postDate": "2020-02-28T01:12:29.133Z",
      "content": "<p><a href=\"/zfturbo\">@zfturbo</a> I was kind of surprised by the fact that this great tool is missed by many people!(Only 2 people put comment?) I want to say thank you, this was really helpful for the task I worked recently.</p>",
      "rawMarkdown": "@zfturbo I was kind of surprised by the fact that this great tool is missed by many people!(Only 2 people put comment?) I want to say thank you, this was really helpful for the task I worked recently.\n",
      "votes": 1
    },
    {
      "id": 941590,
      "postDate": "2020-07-23T10:08:16.523Z",
      "content": "<p>Very promising tool. Can you detail the theory/algorithm behind your contribution?</p>",
      "rawMarkdown": "Very promising tool. Can you detail the theory/algorithm behind your contribution?",
      "replies": [
        {
          "id": 942122,
          "postDate": "2020-07-23T15:39:04.487Z",
          "content": "<p>Some details in here:\n<a href=\"https://arxiv.org/abs/1910.13302\">https://arxiv.org/abs/1910.13302</a></p>\n\n<p>I plan to add more details later.</p>",
          "rawMarkdown": "Some details in here:\nhttps://arxiv.org/abs/1910.13302\n\nI plan to add more details later."
        }
      ]
    },
    {
      "id": 664530,
      "postDate": "2019-11-03T20:04:22.887Z",
      "rawMarkdown": "",
      "isDeleted": true
    },
    {
      "id": 663156,
      "postDate": "2019-11-01T14:21:10.283Z",
      "content": "<p>Wonderful work! Thank you!</p>",
      "rawMarkdown": "Wonderful work! Thank you!"
    }
  ],
  "comments": [
    {
      "id": 758617,
      "author_name": "Camaro",
      "author_url": "",
      "post_date": "2020-02-28T01:12:29.133000",
      "content": "<p><a href=\"/zfturbo\">@zfturbo</a> I was kind of surprised by the fact that this great tool is missed by many people!(Only 2 people put comment?) I want to say thank you, this was really helpful for the task I worked recently.</p>",
      "votes": 1,
      "replies": []
    },
    {
      "id": 941590,
      "author_name": "Haithem Hermassi",
      "author_url": "",
      "post_date": "2020-07-23T10:08:16.523000",
      "content": "<p>Very promising tool. Can you detail the theory/algorithm behind your contribution?</p>",
      "votes": 0,
      "replies": [
        {
          "id": 942122,
          "author_name": "ZFTurbo",
          "author_url": "",
          "post_date": "2020-07-23T15:39:04.487000",
          "content": "<p>Some details in here:\n<a href=\"https://arxiv.org/abs/1910.13302\">https://arxiv.org/abs/1910.13302</a></p>\n\n<p>I plan to add more details later.</p>",
          "votes": 0,
          "replies": []
        }
      ]
    },
    {
      "id": 664530,
      "author_name": "",
      "author_url": "",
      "post_date": "2019-11-03T20:04:22.887000",
      "content": "",
      "votes": 0,
      "replies": []
    },
    {
      "id": 663156,
      "author_name": "Zehao Zhao",
      "author_url": "",
      "post_date": "2019-11-01T14:21:10.283000",
      "content": "<p>Wonderful work! Thank you!</p>",
      "votes": 0,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "662160": "I made a release of Python module for ensembling boxes for Object Detection models. It includes very promicing Weighted Boxes Fusion (WBF) method, which we used instead of NMS in 2019 and 2018 competitions.\n\nhttps://github.com/ZFTurbo/Weighted-Boxes-Fusion\n\nRepository contains the following methods:\n- Non-maximum Suppression (NMS)\n- Soft-NMS\n- Non-maximum weighted (NMW)\n- Weighted boxes fusion (WBF)\n\nInstallation available using pip: `pip install ensemble-boxes`\n\nIt will be interesting to hear if WBF works better for other teams too. )",
    "758617": "@zfturbo I was kind of surprised by the fact that this great tool is missed by many people!(Only 2 people put comment?) I want to say thank you, this was really helpful for the task I worked recently.\n",
    "941590": "Very promising tool. Can you detail the theory/algorithm behind your contribution?",
    "664530": "",
    "663156": "Wonderful work! Thank you!"
  }
}