{
  "id": 613969,
  "title": "Exploring CSIRO Image2Biomass Prediction – Seeking Insights and Collaboration",
  "url": "/competitions/csiro-biomass/discussion/613969",
  "author_name": "Didar-E-Kibria",
  "post_date": "2025-10-31T05:44:09.624000",
  "votes": 4,
  "comment_count": 0,
  "views": 0,
  "content": "<p>Hello everyone,\nI’ve recently started exploring the CSIRO Image2Biomass Prediction project — an exciting application of computer vision and machine learning in estimating forest or crop biomass from satellite imagery.</p>\n<p>I’d like to open a discussion around:</p>\n<p>What feature extraction or image preprocessing techniques have worked best in your experiments?</p>\n<p>How can we handle the spatial variability and spectral noise in biomass prediction models?</p>\n<p>Are there any open datasets or pretrained models (e.g., Sentinel-2, Landsat-based) that you’ve found effective for this task?</p>\n<p>I’m particularly interested in comparing deep learning approaches (CNN, U-Net, or ViT-based) versus traditional regression models.\nWould love to hear your thoughts, code snippets, or even visualization ideas!</p>\n<p>Let’s collaborate and push this topic forward. </p>",
  "messages": [
    {
      "id": 3309173,
      "postDate": "2025-10-31T05:44:09.623Z",
      "content": "<p>Hello everyone,\nI’ve recently started exploring the CSIRO Image2Biomass Prediction project — an exciting application of computer vision and machine learning in estimating forest or crop biomass from satellite imagery.</p>\n<p>I’d like to open a discussion around:</p>\n<p>What feature extraction or image preprocessing techniques have worked best in your experiments?</p>\n<p>How can we handle the spatial variability and spectral noise in biomass prediction models?</p>\n<p>Are there any open datasets or pretrained models (e.g., Sentinel-2, Landsat-based) that you’ve found effective for this task?</p>\n<p>I’m particularly interested in comparing deep learning approaches (CNN, U-Net, or ViT-based) versus traditional regression models.\nWould love to hear your thoughts, code snippets, or even visualization ideas!</p>\n<p>Let’s collaborate and push this topic forward. </p>",
      "rawMarkdown": "Hello everyone,\nI’ve recently started exploring the CSIRO Image2Biomass Prediction project — an exciting application of computer vision and machine learning in estimating forest or crop biomass from satellite imagery.\n\nI’d like to open a discussion around:\n\nWhat feature extraction or image preprocessing techniques have worked best in your experiments?\n\nHow can we handle the spatial variability and spectral noise in biomass prediction models?\n\nAre there any open datasets or pretrained models (e.g., Sentinel-2, Landsat-based) that you’ve found effective for this task?\n\nI’m particularly interested in comparing deep learning approaches (CNN, U-Net, or ViT-based) versus traditional regression models.\nWould love to hear your thoughts, code snippets, or even visualization ideas!\n\nLet’s collaborate and push this topic forward. ",
      "votes": 4
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "3309173": "Hello everyone,\nI’ve recently started exploring the CSIRO Image2Biomass Prediction project — an exciting application of computer vision and machine learning in estimating forest or crop biomass from satellite imagery.\n\nI’d like to open a discussion around:\n\nWhat feature extraction or image preprocessing techniques have worked best in your experiments?\n\nHow can we handle the spatial variability and spectral noise in biomass prediction models?\n\nAre there any open datasets or pretrained models (e.g., Sentinel-2, Landsat-based) that you’ve found effective for this task?\n\nI’m particularly interested in comparing deep learning approaches (CNN, U-Net, or ViT-based) versus traditional regression models.\nWould love to hear your thoughts, code snippets, or even visualization ideas!\n\nLet’s collaborate and push this topic forward. "
  }
}