We have several system solutions for using face recognition. Here is the distributed network solution which separates the roles during face recognition
DISTRIBUTED NETWORK SOLUTION
Works are separated effectively between local and network.
In local side (store or the location of camera installments), faces are detected, gender and age information of people are analyzed, person biometrics template (AFID) is generated and sent to the main server for comparison with database.
In Traditional solutions, whole live camera frames are sent to the server which is 100% server dependent.
In centralized solution, only face image is sent to the server by reducing the number of data to be sent hence a fast matching processing speed with minimized network load.
In semi-centralized solution, face detection, afid generations are done in local side and only person afid data is sent to the server for matching with person face database.
In distributed architecture, everything is done in local side including matching but whole system is observed by a VPN-connected central database server. All backup of the system is taken by central server and in case of any failure at local, central server synchronizes the latest data, informs each local site for this failures. Therefore, system runs 24 hours with auto-recovery features.
ARCHITECTURE

NETWORK DIAGRAM OF THE DISTRIBUTED SOLUTION

ADVANTAGES
- Network-independent operation
- Very-low network load
- Cheap hardware costs
- No data loss in local site due of malfunctions
- Easy location change in local
- Maintenance-free system
WHAT IS IN AND OUT
What you give | Live camera images |
What you get | l Gender informationl Age informationl Happiness information of the personl Face recognition and verification of the person
l Face alarms(Sound, Email, Door signal) |