Connect to a Video Feed

The Desktop Client requires a video feed window to be open for each camera feed (or video file) it is monitoring. The video feed window not only facilitates SAFR monitoring of the video but can also present additional information overlaid on top of the video to assist staff in interpreting the scene in the video. For more information on the additional overlaid information, see Interpret Video Feed Overlays.

The Desktop Client automatically detects integrated, USB, and Open Network Video Interface Forum (ONVIF)-compatible IP cameras. While you’re becoming familiar with SAFR, we recommend that you plug in a single USB camera (aka a web cam). Only after you’ve spent some time learning SAFR should you attempt to connect to additional IP cameras.

The instructions on this page assume you have at least one camera detectable from the computer where SAFR is installed. See Connect Cameras to SAFR for information about how to connect a camera.

Connect to a Camera Video Feed with the Desktop Client

If you have an on-premise deployment, you must first install and configure SAFR Platform before connecting to a live video feed. For information on how to do this, see Getting Started with SAFR Platform on Windows or macOS or Getting Started with SAFR Platform on Linux.

If you have a cloud deployment, your Desktop Client should automatically be configured to connect to the SAFR Server in the cloud.

To connect to a video feed, do the following:

  1. Start your Desktop Client. It should default to the Camera Feed Analyzer window, shown below.

    If it doesn’t default to that window, you can open a new Camera Feed Analyzer window by clicking on Tools > Camera Feed Analyzer….
    Note: In the Operator Console window, the Tools menu is the wrench, as shown below.
  2. Along the top of the Camera Feed Analyzer window, select a camera from the Camera Selector drop-down menu. The menu displays all detected cameras.
  3. Set the camera frame rate and resolution in the Video Feed Profile Selector to the right of the Camera Selector menu. Frame rate and resolution selection are only configurable for USB and IP cameras auto-discovered via ONVIF. For IP ONVIF cameras, only configured ONVIF media profiles are able to be selected. The higher the frame rate (i.e. frames per second) and resolution for a video feed, the more processing power is required to monitor and collect data from the video feed.

Once you complete this procedure, SAFR receives, monitors, and processes the video feed from the camera. The Desktop Client Camera Feed Analyzer window must remain open for SAFR to continue to monitor the video feed. If the window is closed, SAFR no longer receives the video feed and no longer monitors it. The Camera Feed Analyzer window can be minimized without affecting the monitoring of the feed.

Select a Video Processing Mode

A variety of different video processing modes are supported to accommodate different monitoring and security needs, as described in the table below. Each mode can be customized through the Detection, Tracking, Recognition, Events, and User Interface tabs of the Preferences Window.

SAFR Video Processing Mode Description
Recognition
  • This is the default mode typically used for set-up, validation, and experimentation.
  • Only reports enrolled individuals.
  • No events are generated or recorded.
  • Available for both Windows and macOS.
Import
  • Any face that can be clearly seen but is unidentified will be automatically registered.
  • Faces that are already registered are only recognized and do not create additional entries in the Person Directory.
  • Additional different face images (e.g. from different expressions) may be added to the existing faces in the Person Directory if they improve the recognition of the person.
  • No events are recorded.
  • Available for both Windows and macOS.
Learn and Monitor
  • Monitors all person events within view. If a person is not registered in the system, they are added as long as the image meets the specified image quality metric criteria.
  • Allows for automatic saving of recognized persons to the server.
  • Available for Windows.
Similar
  • Registered faces that are the most similar to the face observed in the video are shown in real time in the Camera Feed Analyzer window.
  • Each similar face displays a % match. Anything less than 100% match is not a certain match.
  • No event information or images are recorded.
  • Available for macOS.
Secure Access
  • Secure access uses strict criteria for confirming the identity of the face in the view of the camera.
  • Records events and images for recognized faces.
  • Listens for event replies and displays them on the screen.
  • Typically used for door access control.
  • Events and images are not recorded for unrecognized faces.
  • Available for both Windows and macOS.
Secure Access with RGB Liveness
  • Secure access with RGB Liveness uses strict criteria for confirming the identity of the face in the view of the camera.
  • Records events and images for recognized individuals.
  • Listens for event replies and displays them on the screen.
  • Typically used for door access control where you want to use RGB liveness detection to guard against face spoofing.
  • See Liveness Detection for information about RGB liveness detection.
  • Only available on Windows machines containing an NVidia card that provides GPU.
Secure Access with Smile
  • Uses strict criteria for confirming the identity of the face in the view of the camera. It also looks for transitions in the facial expression (i.e. smile authorization).
  • Records events and images for recognized individuals.
  • Listens for event replies and displays them on the screen.
  • Typically used for door access control where it is necessary to guard against identity impersonation via photo.
  • Includes all functionality of Secure Access mode and can be used to allow access on first sight recognition for a certain time (e.g. when security staff are on duty) and can change to a higher degree of security (e.g. recognition with smile expression change) at set times (e.g. when security staff is off duty).
  • Events and images are not recorded for unrecognized faces.
  • Available for both Windows and macOS.
Secure Access with Smile and RGB Liveness
  • Uses strict criteria for confirming the identity of the face in the view of the camera. It also looks for transitions in the facial expression.
  • Records events and images for recognized individuals.
  • Listens for event replies and displays them on the screen.
  • Typically used for door access control where it is necessary to guard against identity impersonation via photo and where you want to use RGB liveness detection to guard against face spoofing.
  • See Liveness Detection for information about RGB liveness detection.
  • Only available on Windows machines containing an NVidia card that provides GPU.
Secure Access with Profile Pose
  • Secure access with Profile Pose uses strict criteria for confirming the identity of the face in the view of the camera.
  • Records events and images for recognized individuals.
  • Listens for event replies and displays them on the screen.
  • Typically used for door access control where you want to use pose liveness detection to guard against face spoofing.
  • See Liveness Detection for information about pose liveness detection.
  • Available for both Windows and macOS.
Registration Kiosk
  • Allows self-registration into the SAFR system.
  • Optimized to work from a tablet. We recommend the iPad Pro.
  • Can be customized in the User Interface preferences tab to display different prompts, gather additional information, or restrict registration with a minimum age.
  • Records registration and appearance events and associated images of registered persons.
  • Events and images are not recorded for unrecognized faces unless they initiate the registration process.
  • Available for macOS.
Enrolled Monitoring
  • Facial recognition, events, and images are only recorded for registered/recognized faces.
  • Images and events are not recorded for unrecognized faces.
  • Available for both Windows and macOS.
Anonymous Traffic Monitoring
  • When gender and/or age detection is enabled, this metadata is recorded anonymously for faces viewed by cameras.
  • Images and biometric information are not recorded.
  • Available for both Windows and macOS.
Enrolled and Anonymous Traffic Monitoring
  • Recognition events are recorded for registered individuals.
  • When gender and/or age detection is enabled, this metadata is recorded anonymously for faces viewed by cameras.
  • No images are recorded for recognized or unrecognized faces.
  • Available for both Windows and macOS.
Enrolled and Unique Traffic Monitoring
  • Events are recorded for registered and unknown individuals.
  • Any faces that can be clearly seen but are currently unknown are automatically registered.
  • Faces that can’t be seen clearly enough to attempt recognition are reported as unrecognizable.
  • Age information is automatically recorded for all clearly seen faces.
  • Images are recorded for all faces.
  • Available for both Windows and macOS.
Enrolled and Stranger Monitoring
  • Events are recorded for registered and unknown individuals.
  • Faces that are clearly seen but aren’t currently registered are reported as strangers.
  • Faces that can’t be seen clearly enough to attempt recognition are reported as unrecognizable.
  • Images are recorded for all faces.
  • Available for both Windows and macOS.

Recommendations for the Best Video Experience

See Also