SAFR Overview
SAFR is an exceptionally accurate AI-powered facial recognition that provides a new level of visibility and situational awareness for security professionals. You can easily integrate access control peripherals such as cameras, door locks, or alert systems in order to manage access to a location based on people's identities. SAFR runs on a variety of operating systems, including Windows, macOS, Linux, iOS, and Android.
SAFR consists of the following components:
- SAFR Server: Available for Windows, macOS, and Linux. SAFR Server consists of a recognition engine, an event server, and several databases. The databases contain stored face images of enrolled people, the identity information for the stored faces, and recognition events that have been generated by the SAFR system.
SAFR Server runs as several background services that automatically start on system reboot and are kept active by the operating system. They must be running at all times for the SAFR system to be operational. In order to be functional, all other SAFR components must maintain a connection to a SAFR Server. Note that if you're doing a cloud deployment you'll be connecting to a SAFR Server in the cloud that RealNetworks maintains.
- Desktop Client: Available for Windows and macOS. The Desktop Client is one of the primary ways that administrators and operators can interact with the SAFR system. As such, the client can be used to enable camera connectivity, monitor video camera feeds, register users, view recognition events, and more.
- Mobile Client: Available for Android and iOS. The Mobile Client converts a mobile device into a registration kiosk or a recognition panel. Registration kiosks allow people to self-register their face into the Identity Database so they can be approved for access or granted other privileges. Recognition panels enable the mobile device to scan the faces of people that walk by and compare those faces against faces in the Identity Database. Mobile devices set up as recognition panels can also provide visual or audio feedback to the person viewing the mobile device based on actions that a SAFR administrator has configured.
- Web Console: Available on all platforms. The Web Console provides administrators and operators web-based access to the SAFR system. As such, the Web Console can be used to generate analytical reports, monitor video camera feeds, register users, view recognition events, and more.
- Video Recognition Gateway (VIRGO): Available as a standalone download for macOS and Linux. It's also available as part of the SAFR Desktop and SAFR Platform download packages. VIRGO is a daemon system which receives video feeds from one or more cameras and recognizes and tracks faces in those video streams in real time. It generates tracking events and sends those events to an event server. The VIRGO feeds can be controlled either by the command line tool or by the Video Feeds window in the Desktop Client or the Web Console.
- Actions Relay Event Service (ARES): Available as a standalone download for all platforms. ARES is a cross-platform Java application that acts as the event listener that dispatches configured actions in response to events. ARES can provide replies on any event handled by the client that originates an event and is normally installed as a service when either SAFR Platform or SAFR Desktop are installed. It is constantly active and is automatically started by the operating system on power-up.
- SAFR Actions: SAFR Actions is a GUI that facilitates configuring SAFRActions.config. SAFRActions.config is the file that defines all the defined actions for your SAFR System, as well as a couple fields that are used to connect ARES (and SAFR Actions) to your primary SAFR Server, whether that server is local or in the cloud. See Actions for more information about actions in SAFR.
In addition to the SAFR components listed above, SAFR also relies on a couple additional non-SAFR components:
- IP Cameras: As you might expect, Internet Protocol (IP) cameras are absolutely integral to SAFR. Both the Desktop Client and VIRGO automatically detect integrated, USB, and Open Network Video Interface Forum (ONVIF) IP cameras. If an IP camera does not support ONVIF or doesn't have ONVIF enabled, you can still manually add it to the SAFR system as described here.
- Physical access control devices: Door locks, electronic gates, etc. can all be used by SAFR to grant or deny access to people, depending on whether or not they're identified as having the proper authorization.
- Notification systems: Email can be used to discretely notify specified people of various events, while general alarms can be used to alert everybody in the vicinity when unauthorized people attempt to force entry.
- Additional external peripherals: Any device that can be controlled by a computer language or protocol can be incorporated into the SAFR system.
The following download packages are available on the SAFR Download Portal:
- SAFR Platform: Available on Windows, macOS, and Linux. The SAFR Platform installs everything you need to set up a local deployment of SAFR. This downlaod package enables a locally deployed system to be easily deployed on a single computer and afterwards expanded to additional computers as needed. See Getting Started with SAFR Platform on Windows or macOS and Getting Started with SAFR Platform on Linux for more information.
- SAFR Desktop: Available on Windows and macOS. Installs the Desktop Client, SAFR Actions, and one of the VMS extensions. Windows has an additional download variant called SAFR Desktop Lite which has fewer features and lower system requirements. See Getting Started with SAFR Desktop with a Cloud Account for more information.
- SAFR Mobile: Available on Android and iOS. Installs the Mobile Client. When you download SAFR Mobile for Android, you're also offered the SAFR Beam download. SAFR Beam allows you to enable the more secure Lock Task Mode on your Android device. If you don't install SAFR Beam, then Android devices can only enable the less secure Screen Pinning Mode. See Configure an Android Device into Lock Task Mode and Configure an iOS Device into Single App Mode for more information.
- Video Recognition Gateway (VIRGO): Available on Linux and macOS. Installs VIRGO.
- Actions Relay Event Service (ARES): Available on all platforms. Installs ARES.
There are two types of SAFR deployment: cloud and on-premises. Each deployment type requires its own account type; a cloud deployment requires a SAFR Cloud Account, while an on-premises deployment requires a SAFR Local Account. Contact your SAFR Account Manager to obtain either type of account.
When SAFR is deployed as a cloud deployment, all your SAFR components are deployed locally except for the SAFR Server. Your components will connect to a SAFR Server located in the cloud which is operated by RealNetworks, Inc. Using the cloud SAFR Server greatly simplifies deployment and maintenance, but it requires a network connection to the cloud at all times in order to be operational.
A single installation of the Desktop Client can handle about 16 connected cameras, assuming the hosting machine meets the recommended system requirements listed here. More powerful machines may be able to handle more than 16 cameras. Expanding your SAFR system beyond this limitation is fairly easy; simply install additional Desktop Clients onto additional machines.
When SAFR is deployed as an on-premises deployment, all of the SAFR components (including SAFR Server) are installed locally. During installation your SAFR Server will attempt to connect to a SAFR License Server in the cloud to obtain a licence, but after a license has been obtained on-premises deployments do not require a connection to the cloud. Please note that it's possible to obtain a SAFR license without ever connecting your SAFR Server to the cloud; see On-Premises Licensing for details.
A single installation of the SAFR Server can handle about 25 viewed faces at one time, assuming the hosting machine meets the recommended system requirements listed here. Note that for the purposes of server capacity, "25 viewed faces" can mean "25 cameras with 1 face in each camera view" or "1 camera with 25 faces in its camera view", or anything in between. If you want to expand your SAFR system beyond this limitation please see SAFR Server Clusters.