Gait as a Biometric

Facial recognition is becoming increasingly powerful, but this may also change human behaviour. A person who wishes to avoid identification can cover the face, turn away from a camera or deliberately obscure key facial features. Gait offers a different possibility because the way a person walks is shaped by body proportions, joint movement, posture, stride and rhythm. These characteristics are difficult to suppress completely without changing the way one moves.

With advances in video analysis and machine learning, it may become possible to extract gait signatures from ordinary surveillance footage and compare them across time and location. This could be useful in criminal investigations when the face is not visible, helping investigators assess whether individuals seen in different recordings may be the same person. At borders or other controlled environments, gait could also provide an additional biometric signal alongside the face rather than acting as a replacement for it.

The attraction of gait recognition is that it can operate at a distance and does not require a person to cooperate with the camera. But it is also inherently variable. Clothing, footwear, injury, fatigue, carried objects and camera viewpoint can all alter appearance and movement. For that reason, gait should be treated as supporting biometric evidence rather than infallible identification. The larger question is how such systems should be governed if they become capable of recognising people even when they deliberately avoid showing their faces.