{
  "id": 399469,
  "title": "Fbeta-score💥",
  "url": "/competitions/vesuvius-challenge-ink-detection/discussion/399469",
  "author_name": "Wwwwhy",
  "post_date": "2023-04-04T07:56:28.632000",
  "votes": 0,
  "comment_count": 0,
  "views": 0,
  "content": "<pre><code> def Fbeta_score(premask, groundtruth,beta=0.5) :\n    premask = premask.flatten()\n    groundtruth = groundtruth.flatten()\n    seg_inv, gt_inv = torch.logical_not(premask), torch.logical_not(groundtruth)\n    true_pos = torch.sum(torch.logical_and(premask, groundtruth))  # float for division\n    # true_neg = np.logical_and(seg_inv, gt_inv).sum()\n    false_pos = torch.logical_and(premask, gt_inv).sum()\n    false_neg = torch.logical_and(seg_inv, groundtruth).sum()\n    p = true_pos / (true_pos + false_pos + 1e-6)\n    r = true_pos / (true_pos + false_neg + 1e-6)\n    fbeta = (1 + beta ** 2) * p * r / (beta ** 2 * p + r + 1e-6)\n    return fbeta\n</code></pre>",
  "messages": [
    {
      "id": 2208629,
      "postDate": "2023-04-04T07:56:28.633Z",
      "content": "<pre><code> def Fbeta_score(premask, groundtruth,beta=0.5) :\n    premask = premask.flatten()\n    groundtruth = groundtruth.flatten()\n    seg_inv, gt_inv = torch.logical_not(premask), torch.logical_not(groundtruth)\n    true_pos = torch.sum(torch.logical_and(premask, groundtruth))  # float for division\n    # true_neg = np.logical_and(seg_inv, gt_inv).sum()\n    false_pos = torch.logical_and(premask, gt_inv).sum()\n    false_neg = torch.logical_and(seg_inv, groundtruth).sum()\n    p = true_pos / (true_pos + false_pos + 1e-6)\n    r = true_pos / (true_pos + false_neg + 1e-6)\n    fbeta = (1 + beta ** 2) * p * r / (beta ** 2 * p + r + 1e-6)\n    return fbeta\n</code></pre>",
      "rawMarkdown": "``` \n def Fbeta_score(premask, groundtruth,beta=0.5) :\n    premask = premask.flatten()\n    groundtruth = groundtruth.flatten()\n    seg_inv, gt_inv = torch.logical_not(premask), torch.logical_not(groundtruth)\n    true_pos = torch.sum(torch.logical_and(premask, groundtruth))  # float for division\n    # true_neg = np.logical_and(seg_inv, gt_inv).sum()\n    false_pos = torch.logical_and(premask, gt_inv).sum()\n    false_neg = torch.logical_and(seg_inv, groundtruth).sum()\n    p = true_pos / (true_pos + false_pos + 1e-6)\n    r = true_pos / (true_pos + false_neg + 1e-6)\n    fbeta = (1 + beta ** 2) * p * r / (beta ** 2 * p + r + 1e-6)\n    return fbeta\n```"
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "2208629": "``` \n def Fbeta_score(premask, groundtruth,beta=0.5) :\n    premask = premask.flatten()\n    groundtruth = groundtruth.flatten()\n    seg_inv, gt_inv = torch.logical_not(premask), torch.logical_not(groundtruth)\n    true_pos = torch.sum(torch.logical_and(premask, groundtruth))  # float for division\n    # true_neg = np.logical_and(seg_inv, gt_inv).sum()\n    false_pos = torch.logical_and(premask, gt_inv).sum()\n    false_neg = torch.logical_and(seg_inv, groundtruth).sum()\n    p = true_pos / (true_pos + false_pos + 1e-6)\n    r = true_pos / (true_pos + false_neg + 1e-6)\n    fbeta = (1 + beta ** 2) * p * r / (beta ** 2 * p + r + 1e-6)\n    return fbeta\n```"
  }
}