Skip to content

Plumerai Multi-Camera Re-Identification

Plumerai Multi-Camera Re-Identification (ReID) allows for assigning consistent identities to tracked people across multiple cameras, even when they leave the camera view for prolonged periods. It runs at the edge or in the cloud.

ReID identifies a person based on visual appearance cues, such as their clothing. It helps maintain identity continuity when a person is not continuously visible. In contrast to Familiar Face Identification (FFID), ReID does not use facial features.

The technical API documentation can be found here.

ReID extends standard object tracking by matching a person after gaps in observation. A person can be identified even if they:

  • Leave the camera’s field of view and return later
  • Are temporarily occluded (for example, behind another person or an object)
  • Move through areas the camera cannot see
  • Reappear after minutes or hours (depending on configuration and scene conditions)

ReID can also match people across multiple cameras. This lets the system maintain a consistent identity as a person moves between different camera views (for example, from a driveway camera to a doorbell camera).

ReID matches people across gaps in observation, but is not designed to reliably handle substantial changes in appearance or context.

Out-of-scope examples include:

  • Re-identifying a person after they have changed clothing.
  • Distinguishing people with near-identical clothing, such as school or company uniforms.
  • Re-identifying a person across long time gaps, such as the next day.