{
  "id": 123787,
  "title": "Some questions about KITTI",
  "url": "/competitions/pku-autonomous-driving/discussion/123787",
  "author_name": "",
  "post_date": "2019-12-30T11:35:48.461231600Z",
  "votes": 2,
  "comment_count": 5,
  "views": 0,
  "content": "<p>What's AP-E AP-M AP-H in KITTI?</p>",
  "messages": [
    {
      "id": "706437",
      "postDate": "12/30/2019 11:35:48",
      "content": "<p>What's AP-E AP-M AP-H in KITTI?</p>",
      "rawMarkdown": "What's AP-E AP-M AP-H in KITTI?",
      "votes": null
    },
    {
      "id": "706586",
      "postDate": "12/30/2019 15:47:04",
      "content": "<p>Average Precision on easy, moderate and hard boxes and each level is defined as follows:</p>\n\n<p><strong>Easy</strong>: Min. bounding box height: 40 Px, Max. occlusion level: Fully visible, Max. truncation: 15 %\n<strong>Moderate</strong>: Min. bounding box height: 25 Px, Max. occlusion level: Partly occluded, Max. truncation: 30 %\n<strong>Hard</strong>: Min. bounding box height: 25 Px, Max. occlusion level: Difficult to see, Max. truncation: 50 %</p>\n\n<p>Ref: <a href=\"http://www.cvlibs.net/datasets/kitti/eval_object.php?obj_benchmark=3d\">http://www.cvlibs.net/datasets/kitti/eval_object.php?obj_benchmark=3d</a></p>",
      "rawMarkdown": "Average Precision on easy, moderate and hard boxes and each level is defined as follows:\n\n**Easy**: Min. bounding box height: 40 Px, Max. occlusion level: Fully visible, Max. truncation: 15 %\n**Moderate**: Min. bounding box height: 25 Px, Max. occlusion level: Partly occluded, Max. truncation: 30 %\n**Hard**: Min. bounding box height: 25 Px, Max. occlusion level: Difficult to see, Max. truncation: 50 %\n\nRef: http://www.cvlibs.net/datasets/kitti/eval_object.php?obj_benchmark=3d",
      "votes": null
    },
    {
      "id": "706590",
      "postDate": "12/30/2019 15:50:24",
      "content": "<p>👍 thanks</p>",
      "rawMarkdown": "👍 thanks",
      "votes": null
    },
    {
      "id": "707128",
      "postDate": "12/31/2019 09:33:04",
      "content": "<p>Hi, do you know the different between KITTI 3DOP split and KITTI SubCNN split? <a href=\"/selameab\">@selameab</a> </p>",
      "rawMarkdown": "Hi, do you know the different between KITTI 3DOP split and KITTI SubCNN split? @selameab",
      "votes": null
    },
    {
      "id": "707450",
      "postDate": "12/31/2019 21:05:05",
      "content": "<p>The two papers report their results on different validation sets. The ids used in SubCNN can be found <a href=\"https://github.com/tanshen/SubCNN/blob/master/KITTI/kitti_ids_val.txt\">here</a>. I'm not sure if this is an official implementation but check <a href=\"https://github.com/mittalrajat/3D-Bounding-Box-Estimation/blob/master/3dop-implementation/3DOP_code_cuDNNv5/3DOP/data/ImageSets/val.txt\">this</a> out for 3DOP.</p>",
      "rawMarkdown": "The two papers report their results on different validation sets. The ids used in SubCNN can be found [here](https://github.com/tanshen/SubCNN/blob/master/KITTI/kitti_ids_val.txt). I'm not sure if this is an official implementation but check [this](https://github.com/mittalrajat/3D-Bounding-Box-Estimation/blob/master/3dop-implementation/3DOP_code_cuDNNv5/3DOP/data/ImageSets/val.txt) out for 3DOP.",
      "votes": null
    },
    {
      "id": "707533",
      "postDate": "01/01/2020 02:12:09",
      "content": "<p>wonderful😃 </p>",
      "rawMarkdown": "wonderful😃",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 706586,
      "author_name": "selameab",
      "author_url": "",
      "post_date": "12/30/2019 15:47:04",
      "content": "<p>Average Precision on easy, moderate and hard boxes and each level is defined as follows:</p>\n\n<p><strong>Easy</strong>: Min. bounding box height: 40 Px, Max. occlusion level: Fully visible, Max. truncation: 15 %\n<strong>Moderate</strong>: Min. bounding box height: 25 Px, Max. occlusion level: Partly occluded, Max. truncation: 30 %\n<strong>Hard</strong>: Min. bounding box height: 25 Px, Max. occlusion level: Difficult to see, Max. truncation: 50 %</p>\n\n<p>Ref: <a href=\"http://www.cvlibs.net/datasets/kitti/eval_object.php?obj_benchmark=3d\">http://www.cvlibs.net/datasets/kitti/eval_object.php?obj_benchmark=3d</a></p>",
      "votes": null,
      "replies": [
        {
          "id": 706590,
          "author_name": "diegojohnson",
          "author_url": "",
          "post_date": "12/30/2019 15:50:24",
          "content": "<p>👍 thanks</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 707128,
          "author_name": "diegojohnson",
          "author_url": "",
          "post_date": "12/31/2019 09:33:04",
          "content": "<p>Hi, do you know the different between KITTI 3DOP split and KITTI SubCNN split? <a href=\"/selameab\">@selameab</a> </p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 707450,
          "author_name": "selameab",
          "author_url": "",
          "post_date": "12/31/2019 21:05:05",
          "content": "<p>The two papers report their results on different validation sets. The ids used in SubCNN can be found <a href=\"https://github.com/tanshen/SubCNN/blob/master/KITTI/kitti_ids_val.txt\">here</a>. I'm not sure if this is an official implementation but check <a href=\"https://github.com/mittalrajat/3D-Bounding-Box-Estimation/blob/master/3dop-implementation/3DOP_code_cuDNNv5/3DOP/data/ImageSets/val.txt\">this</a> out for 3DOP.</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 707533,
          "author_name": "diegojohnson",
          "author_url": "",
          "post_date": "01/01/2020 02:12:09",
          "content": "<p>wonderful😃 </p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "706437": "What's AP-E AP-M AP-H in KITTI?",
    "706586": "Average Precision on easy, moderate and hard boxes and each level is defined as follows:\n\n**Easy**: Min. bounding box height: 40 Px, Max. occlusion level: Fully visible, Max. truncation: 15 %\n**Moderate**: Min. bounding box height: 25 Px, Max. occlusion level: Partly occluded, Max. truncation: 30 %\n**Hard**: Min. bounding box height: 25 Px, Max. occlusion level: Difficult to see, Max. truncation: 50 %\n\nRef: http://www.cvlibs.net/datasets/kitti/eval_object.php?obj_benchmark=3d",
    "706590": "👍 thanks",
    "707128": "Hi, do you know the different between KITTI 3DOP split and KITTI SubCNN split? @selameab",
    "707450": "The two papers report their results on different validation sets. The ids used in SubCNN can be found [here](https://github.com/tanshen/SubCNN/blob/master/KITTI/kitti_ids_val.txt). I'm not sure if this is an official implementation but check [this](https://github.com/mittalrajat/3D-Bounding-Box-Estimation/blob/master/3dop-implementation/3DOP_code_cuDNNv5/3DOP/data/ImageSets/val.txt) out for 3DOP.",
    "707533": "wonderful😃"
  },
  "source": "meta"
}