Face database

Face database is a part of Persons database, which includes people and their anthropometric data. Face database is used for comparison with faces that Face Tracker/Recognizer module detects on camera image.

Important

The size is displayed in the Face database size field in the Settings -> Face Recognition window and is determined by the licence for TRASSIR.

Tip

For a detailed description of the person creation process, see Persons section.

Important

TRASSIR can use a unified or central face database. For this purpose, a number of requirements should be met:

  • All face detecting servers and the server, containing the central face database, should have the relevant licenses.
  • You should select the server, containing the central face database from the Face recognizer setup in the Location list.
  • Face detecting servers should connect to the central face database server regularly to synchronize the data. In order to reduce the network load, a special script can be used. To receive the script and its description, please, contact technical support.
Recommendations to the photos used for recognition

TRASSIR uses all photos in the Face database for recognition. The probability of recognizing a person caught in a camera depends on the quality of the photos loaded to the TRASSIR. To increase the probability of recognition, use the following guidelines:

  • You can upload several photos for one person, one of which must be taken in full-face, and the rest are allowed to rotate no more than 30 degrees vertically or horizontally.

    Photos in which the person is depicted half-face, will significantly reduce the probability of this person recognition.

  • If a person wears glasses, then upload a photo with glasses to improve recognition.
  • Photos on which a person's face is blurry, lighted or in shadow will significantly reduce the probability of recognition.
  • The face on the photos should not be smaller than 128x128px.

Examples of photos that improve the quality of recognition

Examples of photos that reduce the quality of recognition