{
  "id": 390950,
  "title": "Mediapipe pose data accuracy",
  "url": "/competitions/asl-signs/discussion/390950",
  "author_name": "",
  "post_date": "2023-02-27T23:54:16.657695300Z",
  "votes": 1,
  "comment_count": 1,
  "views": 0,
  "content": "<p>I have worked with mediapipe library and just want to share that it provides the confidence score for key points, but these are not included in the dataset for the competition. If you use mediapipe for other use cases in future, please consider referencing the confidence scores for EDA and possibly even in data to train the model.</p>",
  "messages": [
    {
      "id": "2162021",
      "postDate": "02/27/2023 23:54:16",
      "content": "<p>I have worked with mediapipe library and just want to share that it provides the confidence score for key points, but these are not included in the dataset for the competition. If you use mediapipe for other use cases in future, please consider referencing the confidence scores for EDA and possibly even in data to train the model.</p>",
      "rawMarkdown": "I have worked with mediapipe library and just want to share that it provides the confidence score for key points, but these are not included in the dataset for the competition. If you use mediapipe for other use cases in future, please consider referencing the confidence scores for EDA and possibly even in data to train the model.",
      "votes": null
    },
    {
      "id": "2162193",
      "postDate": "02/28/2023 03:35:52",
      "content": "<p>\"but these are not included in the dataset for the competition. \"</p>\n<p>you probabilty can train another confidence estimator from the landmark as input: point--&gt;confidence<br>\n(for this you need to generate your train data from video--&gt; [point, confidence] )</p>",
      "rawMarkdown": "\"but these are not included in the dataset for the competition. \"\n\nyou probabilty can train another confidence estimator from the landmark as input: point-->confidence\n(for this you need to generate your train data from video--> [point, confidence] )",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 2162193,
      "author_name": "hengck23",
      "author_url": "",
      "post_date": "02/28/2023 03:35:52",
      "content": "<p>\"but these are not included in the dataset for the competition. \"</p>\n<p>you probabilty can train another confidence estimator from the landmark as input: point--&gt;confidence<br>\n(for this you need to generate your train data from video--&gt; [point, confidence] )</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "2162021": "I have worked with mediapipe library and just want to share that it provides the confidence score for key points, but these are not included in the dataset for the competition. If you use mediapipe for other use cases in future, please consider referencing the confidence scores for EDA and possibly even in data to train the model.",
    "2162193": "\"but these are not included in the dataset for the competition. \"\n\nyou probabilty can train another confidence estimator from the landmark as input: point-->confidence\n(for this you need to generate your train data from video--> [point, confidence] )"
  },
  "source": "meta"
}