Metadata is information that describes object-related content in the camera FOV.
In C-Werk, metadata can be obtained in two ways:
Attention!
To extract metadata from video, you have to de-compress and analyze the video stream, which increases the Server's workload, thus limiting the number of available camera channels.
The following tools are used for the Server-side analysis and metadata generation:
Note
Object tracker and Neural Tracker generate metadata containing the following information about moving objects in FOV: object type, position, size and color, motion speed and direction, etc.
VMD.
Note
VMD generates less accurate data. It does not detect object type and color.
Note
Face detection metadata contains face bounding boxes and their positions, as well as face vectors.
Automatic Number Plate Recognition (LPR/ANPR) tools.
Note
ANPR metadata contains license plate bounding boxes and their positions, as well as vehicle registration numbers.
Note
Metadata from pose detection tools contains information on positions and pose (skeleton) of all people in FOV.
Personal protective equipment detection tools.
Note
Equipment detection metadata contains information about the position of all the people in FOV.
The metadata is used for the following system options:
Attention!
If a camera uses several sources of metadata, the required source is selected automatically, except for MomentQuest.
To perform face/license number searches, only metadata from corresponding detection tools is used.
By default, metadata is stored as files in the object trajectory database in the local Server directory that was selected when installing C-Werk (see Installation) in the vmda_db\VMDA_DB.0\vmda_schema subfolder.
Note
If necessary, you can place metadata on any available network storage (see Configuring storage of the system log and metadata).