External face-data complete workflow

This article combines the external face-data helpers into one complete workflow.

These helpers work with externally generated face-analysis CSV files. They do not infer facial expressions from Gazepoint CSV files and do not treat facial behaviour as direct evidence of emotion.

Complete workflow

face_raw <- read_gazepoint_face_export(
  file = 'external_face_export.csv',
  source = 'openface'
)

face_std <- standardize_gazepoint_face_columns(face_raw)

face_quality <- audit_gazepoint_face_quality(face_std)

face_synced <- sync_gazepoint_face_data(
  gazepoint_data = all_gaze,
  face_data = face_std,
  by = c(participant_id = 'participant_id'),
  gaze_time_col = 'time_sec',
  tolerance_sec = 0.050
)

face_sync_qc <- audit_gazepoint_face_sync(face_synced)

face_windows <- summarize_gazepoint_face_windows(
  face_std,
  windows = windows,
  group_cols = c('participant_id', 'trial_id'),
  window_label_col = 'window',
  measure_cols = c('AU04_r', 'AU12_r')
)

face_reactivity <- summarize_gazepoint_face_reactivity(
  face_windows,
  baseline_window = 'baseline',
  response_window = 'response'
)

face_model <- fit_gazepoint_face_window_lmm(
  face_windows,
  outcome = 'rating',
  predictors = c('AU04_r_mean', 'AU12_r_mean')
)

face_report <- report_gazepoint_face_qc(
  face_data = face_std,
  quality_audit = face_quality,
  sync_audit = face_sync_qc,
  window_summary = face_windows,
  reactivity_summary = face_reactivity,
  multimodal_model = face_model
)

Reporting note

Report the external tool, version, exported variables, confidence/validity fields, synchronisation tolerance, window definitions, and model specification. Avoid unsupported claims of true emotion detection or hidden affect.