{
  "id": 613705,
  "title": "Total size/images in Test.csv?",
  "url": "/competitions/csiro-biomass/discussion/613705",
  "author_name": "",
  "post_date": "2025-10-29T03:44:56.069072900Z",
  "votes": 4,
  "comment_count": 5,
  "views": 0,
  "content": "<p>How many images in test.csv?</p>",
  "messages": [
    {
      "id": "3308300",
      "postDate": "10/29/2025 03:44:56",
      "content": "<p>How many images in test.csv?</p>",
      "rawMarkdown": "How many images in test.csv?",
      "votes": null
    },
    {
      "id": "3308305",
      "postDate": "10/29/2025 03:55:28",
      "content": "<p>The paper says the whole benchmark dataset contains over 1,100 samples. Maybe the test data is the remaining portion of the dataset, around 800–900 samples, I guess.</p>",
      "rawMarkdown": "The paper says the whole benchmark dataset contains over 1,100 samples. Maybe the test data is the remaining portion of the dataset, around 800–900 samples, I guess.",
      "votes": null
    },
    {
      "id": "3308307",
      "postDate": "10/29/2025 04:02:00",
      "content": "<p>Thanks <a href=\"https://www.kaggle.com/lizhecheng\" target=\"_blank\">@lizhecheng</a>, as per paper: <a href=\"https://arxiv.org/pdf/2510.22916\" target=\"_blank\">https://arxiv.org/pdf/2510.22916</a> -&gt; \n1,162 images</p>\n<table>\n<thead>\n<tr>\n<th>Images</th>\n<th>Split</th>\n</tr>\n</thead>\n<tbody>\n<tr>\n<td><strong>1,162</strong></td>\n<td>Paper</td>\n</tr>\n<tr>\n<td><strong>357</strong></td>\n<td>Train</td>\n</tr>\n<tr>\n<td><strong>805</strong></td>\n<td>LB</td>\n</tr>\n<tr>\n<td><strong>425</strong></td>\n<td>Pub LB</td>\n</tr>\n<tr>\n<td><strong>380</strong></td>\n<td>Prv LB</td>\n</tr>\n</tbody>\n</table>\n<hr>\n<p>=&gt; Any external similar datasets available? </p>",
      "rawMarkdown": "Thanks @lizhecheng, as per paper: [https://arxiv.org/pdf/2510.22916](https://arxiv.org/pdf/2510.22916) -> \n1,162 images\n| Images | Split  |\n| --- | --- |\n| **1,162** | Paper |\n|  **357** | Train |\n|  **805** | LB | \n|  **425** | Pub LB |\n| **380** | Prv LB |\n\n-----\n\n=> Any external similar datasets available?",
      "votes": null
    },
    {
      "id": "3308309",
      "postDate": "10/29/2025 04:05:49",
      "content": "<p>Seems not, only 357 training samples; feels like it’s easy to overfit on the training set and LB.</p>",
      "rawMarkdown": "Seems not, only 357 training samples; feels like it’s easy to overfit on the training set and LB.",
      "votes": null
    },
    {
      "id": "3308332",
      "postDate": "10/29/2025 05:54:57",
      "content": "<p>But since the LB only has about 53% of the images, the total number of images in the LB and PB combined is actually about the same as in the training set.</p>",
      "rawMarkdown": "But since the LB only has about 53% of the images, the total number of images in the LB and PB combined is actually about the same as in the training set.",
      "votes": null
    },
    {
      "id": "3309094",
      "postDate": "10/31/2025 00:03:14",
      "content": "<p>The Data tab indicates </p>\n<p>\"The test set contains over 800 images.\"</p>",
      "rawMarkdown": "The Data tab indicates \n\n\"The test set contains over 800 images.\"",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 3308305,
      "author_name": "lizhecheng",
      "author_url": "",
      "post_date": "10/29/2025 03:55:28",
      "content": "<p>The paper says the whole benchmark dataset contains over 1,100 samples. Maybe the test data is the remaining portion of the dataset, around 800–900 samples, I guess.</p>",
      "votes": null,
      "replies": [
        {
          "id": 3308307,
          "author_name": "seshurajup",
          "author_url": "",
          "post_date": "10/29/2025 04:02:00",
          "content": "<p>Thanks <a href=\"https://www.kaggle.com/lizhecheng\" target=\"_blank\">@lizhecheng</a>, as per paper: <a href=\"https://arxiv.org/pdf/2510.22916\" target=\"_blank\">https://arxiv.org/pdf/2510.22916</a> -&gt; \n1,162 images</p>\n<table>\n<thead>\n<tr>\n<th>Images</th>\n<th>Split</th>\n</tr>\n</thead>\n<tbody>\n<tr>\n<td><strong>1,162</strong></td>\n<td>Paper</td>\n</tr>\n<tr>\n<td><strong>357</strong></td>\n<td>Train</td>\n</tr>\n<tr>\n<td><strong>805</strong></td>\n<td>LB</td>\n</tr>\n<tr>\n<td><strong>425</strong></td>\n<td>Pub LB</td>\n</tr>\n<tr>\n<td><strong>380</strong></td>\n<td>Prv LB</td>\n</tr>\n</tbody>\n</table>\n<hr>\n<p>=&gt; Any external similar datasets available? </p>",
          "votes": null,
          "replies": [
            {
              "id": 3308309,
              "author_name": "lizhecheng",
              "author_url": "",
              "post_date": "10/29/2025 04:05:49",
              "content": "<p>Seems not, only 357 training samples; feels like it’s easy to overfit on the training set and LB.</p>",
              "votes": null,
              "replies": [
                {
                  "id": 3308332,
                  "author_name": "gunianxue",
                  "author_url": "",
                  "post_date": "10/29/2025 05:54:57",
                  "content": "<p>But since the LB only has about 53% of the images, the total number of images in the LB and PB combined is actually about the same as in the training set.</p>",
                  "votes": null,
                  "replies": []
                }
              ]
            }
          ]
        }
      ]
    },
    {
      "id": 3309094,
      "author_name": "pcjimmmy",
      "author_url": "",
      "post_date": "10/31/2025 00:03:14",
      "content": "<p>The Data tab indicates </p>\n<p>\"The test set contains over 800 images.\"</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "3308300": "How many images in test.csv?",
    "3308305": "The paper says the whole benchmark dataset contains over 1,100 samples. Maybe the test data is the remaining portion of the dataset, around 800–900 samples, I guess.",
    "3308307": "Thanks @lizhecheng, as per paper: [https://arxiv.org/pdf/2510.22916](https://arxiv.org/pdf/2510.22916) -> \n1,162 images\n| Images | Split  |\n| --- | --- |\n| **1,162** | Paper |\n|  **357** | Train |\n|  **805** | LB | \n|  **425** | Pub LB |\n| **380** | Prv LB |\n\n-----\n\n=> Any external similar datasets available?",
    "3308309": "Seems not, only 357 training samples; feels like it’s easy to overfit on the training set and LB.",
    "3308332": "But since the LB only has about 53% of the images, the total number of images in the LB and PB combined is actually about the same as in the training set.",
    "3309094": "The Data tab indicates \n\n\"The test set contains over 800 images.\""
  },
  "source": "meta"
}