Mark on a real video when each passenger stands up, steps into the aisle, takes their bag down and leaves the plane. calibrate.py turns the resulting CSV into model parameters. The video plays in your browser and is never uploaded.
The time factor is 1 for a normal video; for a time-lapse, the real seconds that each second of video covers.
Space play or pause · ←→ 1 s · Shift+arrow 5 s · ,. one frame · Backspace removes the last mark ·
Green: has some marks. Grey: already off the plane. There is no need to annotate everyone; following a few people carefully is enough.
Samples per parameter; calibrate.py needs at least 8, across all flights.
| Real time | Seat | Event | Value |
|---|
Copy the CSV and save it as name.csv; then run python calibrate.py name.csv --out calibrated.json. You can also paste an earlier CSV here and load it to carry on.
Privacy: do not publish videos in which passengers can be recognised. The CSV only contains times and seat numbers, so it is anonymous.