Baseline calibration captures an individual's neutral resting state at the start of a session, then measures all subsequent readings as deviation from that personal baseline rather than against a population average.
What it is
Faces differ enormously at rest. Someone with naturally downturned lip corners will register as mildly sad against a population norm, permanently and incorrectly. The same applies to behaviour: a person who habitually works with a dozen tabs open looks disengaged against any fixed threshold.
Calibration solves this by establishing what neutral looks like for this person, usually across the first N frames or the first few minutes, then reporting deviation from that.
Why it matters
It is the difference between a system that measures people and one that measures change. Deviation-from-own-baseline is far more defensible statistically and far less discriminatory in practice, because it does not encode the majority's resting state as the standard.
A common confusion
Calibration fixes the individual-difference problem. It does not fix the deeper question of whether facial configurations map reliably to internal states at all.
Related
See also facial action coding system (facs), emotion inference and engagement index. See the full glossary for the rest, or the limits of facial emotion recognition for the wider context.
See where your content loses people
Book a walkthrough and we will show you the engagement data on your own content.