Face recognition in the city video surveillance system is a function for informing law enforcement agencies about the detected match of a passerby’s face with an image of a wanted person. The tool is implemented using computer vision algorithms.

The system consists of two main parts: a detector and video analytics algorithms. The detector highlights the faces of people in the frames, and then video analytics algorithms convert the selected images into a unique digital code and compare it with the digital codes from the database. Images from the wanted database are provided by law enforcement officers. If the system finds a match, it will send a signal.

Moscow’s face recognition system is based on the multi-vendor model: four video analytics algorithms from various vendors are connected to a single detector. This allows to combine the advantages of each developer, and achieve higher recognition accuracy.

Recordings from the cameras are just images and do not contain personal data of residents. Video analytics does not identify a citizen - it only determines how much the image from the camera matches the ‘reference’ image from the database.