{
  "id": 596533,
  "title": "What is the sampling frequency of the triaxial acceleration and quaternions provided in the competition? ",
  "url": "/competitions/cmi-detect-behavior-with-sensor-data/discussion/596533",
  "author_name": "",
  "post_date": "2025-08-04T08:17:44.403819300Z",
  "votes": 1,
  "comment_count": 3,
  "views": 0,
  "content": "<p>When I want to extract some frequency domain features, I'm unsure what to set the sampling frequency to.</p>",
  "messages": [
    {
      "id": "3262753",
      "postDate": "08/04/2025 08:17:44",
      "content": "<p>When I want to extract some frequency domain features, I'm unsure what to set the sampling frequency to.</p>",
      "rawMarkdown": "When I want to extract some frequency domain features, I'm unsure what to set the sampling frequency to.",
      "votes": null
    },
    {
      "id": "3263644",
      "postDate": "08/05/2025 13:58:00",
      "content": "<p>The <a href=\"https://fcon_1000.projects.nitrc.org/indi/cmi_healthy_brain_network/Competitions/Helios2025/IMU_Sensor.pdf\" target=\"_blank\">document</a> provided by the host states that the sampling rate is 1KHz.</p>",
      "rawMarkdown": "The [document](https://fcon_1000.projects.nitrc.org/indi/cmi_healthy_brain_network/Competitions/Helios2025/IMU_Sensor.pdf) provided by the host states that the sampling rate is 1KHz.",
      "votes": null
    },
    {
      "id": "3263883",
      "postDate": "08/05/2025 20:36:45",
      "content": "<p>However, the data seems to be heavily downsampled to ~10Hz<br>\nThis assumption can be justified by seeing that the example videos are all around ~4seconds long while most of the gesture phases are around ~40 samples (indicating 10Hz)</p>",
      "rawMarkdown": "However, the data seems to be heavily downsampled to ~10Hz\nThis assumption can be justified by seeing that the example videos are all around ~4seconds long while most of the gesture phases are around ~40 samples (indicating 10Hz)",
      "votes": null
    },
    {
      "id": "3263976",
      "postDate": "08/06/2025 01:18:55",
      "content": "<p>Good point! But if you set the same SR in training and inference, ML models might handle that automatically.</p>",
      "rawMarkdown": "Good point! But if you set the same SR in training and inference, ML models might handle that automatically.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 3263644,
      "author_name": "atl132",
      "author_url": "",
      "post_date": "08/05/2025 13:58:00",
      "content": "<p>The <a href=\"https://fcon_1000.projects.nitrc.org/indi/cmi_healthy_brain_network/Competitions/Helios2025/IMU_Sensor.pdf\" target=\"_blank\">document</a> provided by the host states that the sampling rate is 1KHz.</p>",
      "votes": null,
      "replies": [
        {
          "id": 3263883,
          "author_name": "patrickaiforfun",
          "author_url": "",
          "post_date": "08/05/2025 20:36:45",
          "content": "<p>However, the data seems to be heavily downsampled to ~10Hz<br>\nThis assumption can be justified by seeing that the example videos are all around ~4seconds long while most of the gesture phases are around ~40 samples (indicating 10Hz)</p>",
          "votes": null,
          "replies": [
            {
              "id": 3263976,
              "author_name": "atl132",
              "author_url": "",
              "post_date": "08/06/2025 01:18:55",
              "content": "<p>Good point! But if you set the same SR in training and inference, ML models might handle that automatically.</p>",
              "votes": null,
              "replies": []
            }
          ]
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "3262753": "When I want to extract some frequency domain features, I'm unsure what to set the sampling frequency to.",
    "3263644": "The [document](https://fcon_1000.projects.nitrc.org/indi/cmi_healthy_brain_network/Competitions/Helios2025/IMU_Sensor.pdf) provided by the host states that the sampling rate is 1KHz.",
    "3263883": "However, the data seems to be heavily downsampled to ~10Hz\nThis assumption can be justified by seeing that the example videos are all around ~4seconds long while most of the gesture phases are around ~40 samples (indicating 10Hz)",
    "3263976": "Good point! But if you set the same SR in training and inference, ML models might handle that automatically."
  },
  "source": "meta"
}