Integrated SAFR Digifort is only available on Windows.
Integrating SAFR's facial recognition and analysis capabilities into Digifort enables you to use SAFR's video feed information overlays within Digifort camera video feeds, thus making it much easier to quickly and accurately separate unknown people from authorized people from known threats. You'll also have immediate access to additional infomation such as age, gender, sentiment, name, company, known associates, or any other configurable information you want to create.
Integrating the two systems also allows SAFR's information about individuals to trigger Digifort alerts and other actions within the Digifort system. Digifort's metadata within bookmarks are enriched with SAFR's additional information, allowing you to more easily find relevant bookmarks.
The SAFR Digifort Integration Guide page discusses how to install and configure SAFR and Digifort so that the two systems are properly integrated. This page discusses how to operate the integrated systems.
If the Digifort events are disabled (gray; no matching Digifort event), refer to Digifort documentation to create Digifort events. Once the events are created, open the Digifort Advanced Settings dialog and, from the menu for the associated SAFR event, select the Digifort event. Once the Digifort events are created and matched, the events are enabled (black). For information on creating events, refer to Digifort documentation.
This affects the titles given to bookmarks created from the respective events.
Note: Each event type is only created if enabled in the SAFR Digifort Preferences tab.
In the SAFR Desktop Client, view the video feed for the camera selected from the camera selection menu. The menu shows the cameras available from the Digifort servers. To enable the row at the bottom of the screen that isolates individual faces, click View > Detection List.
Digifort creates bookmarks to help locate important events. Bookmarks are populated with Person Type, ID Class, and Name. They can also provide more detailed information with even more search attributes, such as Age and Gender.
The following illustration shows how bookmarks can be used to review important events, such as the detection of a stranger tailgating behind a registered user.
To view Digifort Bookmarks, do the following:
To add people through the SAFR Desktop Client from an image or video file, do the following:
Warning: Reducing detection and recognition settings lowers the quality of the reference face and negatively impacts recognition. It is preferable to increase the quality of your sources than to lower the criteria for learning.
Warning: Users added to SAFR are not synchronized with Digifort; these users exist only in SAFR.
It may be desirable to edit people properties to control which events get triggered when that person is recognized. For example, setting a person's ID Class to Concern or Threat triggers the respective alarms. The most important people attributes are Name, Image, Person Type, and ID Class.
Name, Image, and Person Type should be edited through SAFR. Person Type defines a person's role (e.g. staff or visitor), while the ID Class defines the risk level (No-Concern, Concern, or Threat). Person Type and Image can be edited in the Desktop Client by changing the Person Type on the People screen.
ID Class and all other attributes of a person are also edited within SAFR People dialog, accessed through the SAFR Desktop Client Tools menu. All identities are created by default with an ID Class of No Concern. To edit a person's ID Class, open the People window from the SAFR Desktop Client Tools menu as follows:
The Person Type and Name can be edited by clicking the respective fields on the People screen. To edit ID Class, double-click the person, and choose the desired value from the ID Class menu in the People Edit dialog as shown in the following graphic:
You can use SAFR to generate events. When enabled on the Events Preferences page, events occur when a person appears in the view of a connected camera. When a person appears in the camera's view, they're immediately assigned an ID Class attribute, although that ID Class may change if the system successfully recognizes them and assigns them a more appropriate ID Class. The types of ID Classes are listed below.
The following table lists the available events that SAFR makes available to Digifort.
Event Message | Id Class | Named | Person Type | Condition | People Attributes |
---|---|---|---|---|---|
Unrecognizable face detected | N/A | N/A | N/A | Face detected but insufficient information for recognition | idClass="unidentified" |
Stranger detected | Stranger | N/A | N/A | Face detected but not found in registered people | idClass="stranger" |
Registered person detected without name | Normal | No | None | Registered person without name or person type assigned | idClass="noconcern" && personType="" && name="" |
Registered person detected with name <name> | Normal | Yes | None | Registered person with name but no person type | idClass="noconcern" && personType="" && name=<name> |
Registered person detected of type <personType> | Normal | No | Defined | Registered person with person type but no name | idClass="noconcern" && personType=<personType> && name="" |
Registered person detected of type <personType> with name <name> | Normal | Yes | Defined | Registered person with person type and name | idClass="noconcern" && personType=<personType> && name=<name> |
Concern person detected without a name | Concern | No | None | Same as above for Concern | idClass="concern" && personType="" && name="" |
Concern person detected with name <name> | Concern | Yes | None | Same as above for Concern | idClass="concern" && personType="" && name=<name> |
Concern person detected of type <personType> | Concern | No | Defined | Same as above for Concern | idClass="concern" && personType=<personType> && name="" |
Concern person detected of type <personType> with name <name> | Concern | Yes | Defined | Same as above for Concern | idClass="concern" && personType=<personType> && name=<name> |
Threat person detected without a name | Threat | No | None | Same as above for Threat | idClass="threat" && personType="" && name="" |
Threat person detected with name <name> | Threat | Yes | None | Same as above for Threat | idClass="threat" && personType="" && name=<name> |
Threat person detected of type <personType> | Threat | No | Defined | Same as above for Threat | idClass="threat" && personType=<personType> && name="" |
Threat person detected of type <personType> with name <name> | Threat | Yes | Defined | Same as above for Threat | idClass="threat" && personType=<personType> && name=<name> |
SAFR has different operator modes that control what events are generated. Below is a short summary of the modes most relevant to Digifort integration. For a complete description, see the Operator Modes documentation.
You can also use SAFR to view recognition events. Recognition events occur when a known, unknown, or unrecognized person appears in the view of a camera. The types of recognized persons are:
There are several different combinations of these conditions that are triggered. The following graphic shows multiple events populated in the Digifort alerts panel:
Note: When closing SAFR, use the Quit SAFR option on the File menu. Closing SAFR using the Window Close button will cause you to lose the SAFR state settings and connected cameras for that window.