Face Recognition Introduction

Face Recognition (FR) is used to identify a human face. It is used in the VMS System to get events when faces are detected from selected video streams and to detect when specific persons are seen in the video. Together with the Mirasys List Management, this allows you to, for example, create an automatic detection system of a person’s access to the premises. Face Recognition and List Management include face tracking and a face similarity value to match a face to an identity. This is done to prevent excessive face detection events of the same person. The maximum number of faces that can be detected from one image in FR is 20.

 

Note that anti-spoofing is not included in the current version.

 

The FR service receives video streams, processes images, detects faces, and sends notifications with detection data to List Management (LM) service for identity and list matching.

Face Recognition service has a separate installer, so it can execute on a separate server or on some VMS server.

Face recognition works with 168x256 image size. If the face is larger in the picture, it is first reduced to 168x256 before identification is made. Likewise, if the face is smaller in the picture, it is first enlarged to the 168x256 size. The recommended size of the face in the image is at least 112 pixels.

On licensing side there is need to add RTSP Streaming Server feature to Management Server and all those servers where is plan to use SmartFR feature. On Management Server side there is need to be add wanted amount of channels of SmartFR feature.