Data collected in landmarker json file

[iainbryson]

✅ There are no timestamps in the json file. (High Importance)

• Time is not in order
• Missing data; is it real or do we send null data
• Last datapoint is messed up
• faceLandmarks visibility field seems to be 0 all the time? (See attached json file)
• vacialTransformationMatrixes is empty, is that because the <output_facial_transformation_matrixes> was set to false?
• check the configuration.
• faceBlendshapes.categories have a field that is called displayName that seems to be empty all the time? Maybe redundant and can be removed?
• headIndex seems to be empty/constant?
• headName seems to be empty/constant?
• clean up unused fields.

✅ We get an error when we try to read the larger json file,

This one

{'id': 1676,
  'stage_id': 2519,
  'study_definition_id': 918,
  'protocol_definition_id': 917,
  'phase_definition_id': 1037,
  'component_id': None,
  'file': "{'url': '/uploads/study_result/time_series/file/1676/face_landmark.json'}",
  'schema': 'face_landmark_json',
  'schema_metadata': None,
  'created_at': <pyswagger.primitives._time.Datetime at 0x1b9f2869820>,
  'updated_at': <pyswagger.primitives._time.Datetime at 0x1b9f2869850>},

Is there a more robust way to save the JSON file?

Ways in which we could reduce errors could be to save Blendshapes and Landmarks in separate files? So face_landmark.json and/or face_blendshape.json

Or maybe there is another data format we could use to make it more robust?

For now that is the feedback. Happy to jump on a call if needed.

Best

Jens

[iainbryson]

I've tried reproducing the issue with the last row/datapoint that you mentioned but cannot. My script to process the landmarker timeseries is here: https://github.com/elicit-experiment/client-api/blob/e1f1a8719b1675238c2e50c1e5bd2b86fdf9f871/examples/parse_facelandmarker_json.py What breaks for you @madjens ?