{
  "id": 410949,
  "title": "Data Augmentation",
  "url": "/competitions/image-matching-challenge-2023/discussion/410949",
  "author_name": "",
  "post_date": "2023-05-17T05:38:33.429472400Z",
  "votes": 4,
  "comment_count": 3,
  "views": 0,
  "content": "<p>If you have control over the picture capture process, please follow these guidelines for optimal reconstruction results:</p>\n<p>Capture images with good texture. Avoid completely texture-less images (e.g., a white wall or empty desk). If the scene does not contain enough texture itself, you could place additional background objects, such as posters, etc.</p>\n<p>Capture images at similar illumination conditions. Avoid high dynamic range scenes (e.g., pictures against the sun with shadows or pictures through doors/windows). Avoid specularities on shiny surfaces.</p>\n<p>Capture images with high visual overlap. Make sure that each object is seen in at least 3 images – the more images the better.</p>\n<p>Capture images from different viewpoints. Do not take images from the same location by only rotating the camera, e.g., make a few steps after each shot. At the same time, try to have enough images from a relatively similar viewpoint. Note that more images is not necessarily better and might lead to a slow reconstruction process. If you use a video as input, consider down-sampling the frame rate.</p>\n<p>I hope it would be useful in Data preprocessing</p>\n<p><a href=\"https://colmap.github.io/tutorial.html\" target=\"_blank\">https://colmap.github.io/tutorial.html</a></p>",
  "messages": [
    {
      "id": "2262721",
      "postDate": "05/17/2023 05:38:33",
      "content": "<p>If you have control over the picture capture process, please follow these guidelines for optimal reconstruction results:</p>\n<p>Capture images with good texture. Avoid completely texture-less images (e.g., a white wall or empty desk). If the scene does not contain enough texture itself, you could place additional background objects, such as posters, etc.</p>\n<p>Capture images at similar illumination conditions. Avoid high dynamic range scenes (e.g., pictures against the sun with shadows or pictures through doors/windows). Avoid specularities on shiny surfaces.</p>\n<p>Capture images with high visual overlap. Make sure that each object is seen in at least 3 images – the more images the better.</p>\n<p>Capture images from different viewpoints. Do not take images from the same location by only rotating the camera, e.g., make a few steps after each shot. At the same time, try to have enough images from a relatively similar viewpoint. Note that more images is not necessarily better and might lead to a slow reconstruction process. If you use a video as input, consider down-sampling the frame rate.</p>\n<p>I hope it would be useful in Data preprocessing</p>\n<p><a href=\"https://colmap.github.io/tutorial.html\" target=\"_blank\">https://colmap.github.io/tutorial.html</a></p>",
      "rawMarkdown": "If you have control over the picture capture process, please follow these guidelines for optimal reconstruction results:\n\nCapture images with good texture. Avoid completely texture-less images (e.g., a white wall or empty desk). If the scene does not contain enough texture itself, you could place additional background objects, such as posters, etc.\n\nCapture images at similar illumination conditions. Avoid high dynamic range scenes (e.g., pictures against the sun with shadows or pictures through doors/windows). Avoid specularities on shiny surfaces.\n\nCapture images with high visual overlap. Make sure that each object is seen in at least 3 images – the more images the better.\n\nCapture images from different viewpoints. Do not take images from the same location by only rotating the camera, e.g., make a few steps after each shot. At the same time, try to have enough images from a relatively similar viewpoint. Note that more images is not necessarily better and might lead to a slow reconstruction process. If you use a video as input, consider down-sampling the frame rate.\n\n\nI hope it would be useful in Data preprocessing\n\nhttps://colmap.github.io/tutorial.html",
      "votes": null
    },
    {
      "id": "2262915",
      "postDate": "05/17/2023 08:16:38",
      "content": "<p>thanks, interesting :) </p>",
      "rawMarkdown": "thanks, interesting :)",
      "votes": null
    },
    {
      "id": "2262926",
      "postDate": "05/17/2023 08:26:01",
      "content": "<p>However, in this competition, attention may need to be paid when using methods such as spatial flipping for augmentation.</p>",
      "rawMarkdown": "However, in this competition, attention may need to be paid when using methods such as spatial flipping for augmentation.",
      "votes": null
    },
    {
      "id": "2263006",
      "postDate": "05/17/2023 08:59:41",
      "content": "<p>Thank you for sharing the guidelines for optimal image capture in the context of reconstruction.</p>",
      "rawMarkdown": "Thank you for sharing the guidelines for optimal image capture in the context of reconstruction.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 2262915,
      "author_name": "beubeu4l",
      "author_url": "",
      "post_date": "05/17/2023 08:16:38",
      "content": "<p>thanks, interesting :) </p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 2262926,
      "author_name": "dwchen",
      "author_url": "",
      "post_date": "05/17/2023 08:26:01",
      "content": "<p>However, in this competition, attention may need to be paid when using methods such as spatial flipping for augmentation.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 2263006,
      "author_name": "ericka42",
      "author_url": "",
      "post_date": "05/17/2023 08:59:41",
      "content": "<p>Thank you for sharing the guidelines for optimal image capture in the context of reconstruction.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "2262721": "If you have control over the picture capture process, please follow these guidelines for optimal reconstruction results:\n\nCapture images with good texture. Avoid completely texture-less images (e.g., a white wall or empty desk). If the scene does not contain enough texture itself, you could place additional background objects, such as posters, etc.\n\nCapture images at similar illumination conditions. Avoid high dynamic range scenes (e.g., pictures against the sun with shadows or pictures through doors/windows). Avoid specularities on shiny surfaces.\n\nCapture images with high visual overlap. Make sure that each object is seen in at least 3 images – the more images the better.\n\nCapture images from different viewpoints. Do not take images from the same location by only rotating the camera, e.g., make a few steps after each shot. At the same time, try to have enough images from a relatively similar viewpoint. Note that more images is not necessarily better and might lead to a slow reconstruction process. If you use a video as input, consider down-sampling the frame rate.\n\n\nI hope it would be useful in Data preprocessing\n\nhttps://colmap.github.io/tutorial.html",
    "2262915": "thanks, interesting :)",
    "2262926": "However, in this competition, attention may need to be paid when using methods such as spatial flipping for augmentation.",
    "2263006": "Thank you for sharing the guidelines for optimal image capture in the context of reconstruction."
  },
  "source": "meta"
}