create bi-directional converter from Lighting Pose and Label Studio.

[robert-s-lee]

Please review Label Studio Setup Instructions

A method to convert Lightning Pose Annotation data with Label Studio Annotation

Lightning Pose Annotation uses:

• ToyMouseRunningData pngs
• the pngs are annotated in CSV

Label Studio can import pre-annotated data JSON-MIN version

[
  {
    "img": "/data/upload/1/18928a62-img1.png",
    "id": 1,
    "kp-1": [
      {
        "x": 96.71717171717172,
        "y": 7.389162561576355,
        "width": 0.5050505050505051,
        "keypointlabels": [
          "Nose"
        ],
        "original_width": 396,
        "original_height": 406
      },
      {
        "x": 92.17171717171718,
        "y": 5.41871921182266,
        "width": 0.5050505050505051,
        "keypointlabels": [
          "Face"
        ],
        "original_width": 396,
        "original_height": 406
      }
    ],
    "annotator": 1,
    "annotation_id": 1,
    "created_at": "2022-07-06T12:49:47.101659Z",
    "updated_at": "2022-07-06T12:49:47.101700Z",
    "lead_time": 5.507
  }
]

TODO:

• [ ] convert toy_datasets/toymouseRunningData/CollectedData_.csv to label studio pre-annotated format for import

@kathleenislee https://github.com/robert-s-lee/lightning-pose/blob/label-studio/tests/utils/csv_to_label_studio.py has sample code to help get started that reads CSV. the script needs to export in JSON-MIN format.

• [ ] convert label studio pre-annotated JSON to toy_datasets/toymouseRunningData/CollectedData_.csv format