The OTS algorithm works by comparing the global movement in the scene with the one produced by an average human being in that scene.

A few sets of events might prevent the OTS algorithm from correctly estimating the human trafic.

The OTS estimate might come out too low:

  • If a scene changes significantly such as caused by a camera movement or if the camera view was obstructed for a significant period. The system needs at least one hour of constant scenery.
  • If there are too few viewers to compute a model of the scenery. Typically, VidiReports needs around 100 viewers detected so that the system can determine the average person size and movement path from which OTS numbers are generated.
  • If the area being observed is too dark. In this case, the system cannot differentiate between people and the surrounding architecture, thus OTS data is too unreliable to be captured.

Inversely, the OTS estimate might be too optimistic if some moving objects in the field of view cause frequent movement: sliding doors, escalator, TV screen, remote vehicle traffic, floating balloons...