{
  "id": 666205,
  "title": "Real-world images for Hyperparameter tuning",
  "url": "/competitions/duality-ai-lunate-ai-geospatial-object-detection/discussion/666205",
  "author_name": "RebekahDuality",
  "post_date": "2026-01-05T20:15:27.034000",
  "votes": 1,
  "comment_count": 0,
  "views": 0,
  "content": "<p>In response to a question we received from a participant, we want to clarify thatyou  are not allowed to use real-world images for the training or testing, BUT you may use real-world images for hyperparameter tuning. \nAgain, you still need to use ONLY synthetic data for the training.</p>",
  "messages": [
    {
      "id": 3386777,
      "postDate": "2026-01-05T20:15:27.033Z",
      "content": "<p>In response to a question we received from a participant, we want to clarify thatyou  are not allowed to use real-world images for the training or testing, BUT you may use real-world images for hyperparameter tuning. \nAgain, you still need to use ONLY synthetic data for the training.</p>",
      "rawMarkdown": "In response to a question we received from a participant, we want to clarify thatyou  are not allowed to use real-world images for the training or testing, BUT you may use real-world images for hyperparameter tuning. \nAgain, you still need to use ONLY synthetic data for the training.",
      "votes": 1
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "3386777": "In response to a question we received from a participant, we want to clarify thatyou  are not allowed to use real-world images for the training or testing, BUT you may use real-world images for hyperparameter tuning. \nAgain, you still need to use ONLY synthetic data for the training."
  }
}