The SAFR Recognition Quality Calculator provides a means to choose the right camera for facial recognition in your site by identifying the recognition quaity for a given set of environmental and camera attributes. This should be used as part of the Site Survey to identify the camera specifications needed for achieving desired results with either recognition or learning of faces with SAFR.
https://docs.real.com/support/tools/recognition_quality_calculator
The SAFR Recognition Quality Calculator is an interactive tool. Its usage is described in this article. The tool can be downloaded here.
You start out by entering the camera resolution and horizontal angle of view (hAOV). This is information that is typically available in the camera specifications page.
If your camera specifications only provide focal length (i.e. 12mm), then you can convert from that to hAOV using the section of the tool:
Once you enter the camera resolution and hAOV, the color coded chart below enables you to read off the distance at which recognition and learning will perform well (the green region).
Three different metrics are provided:
- Angle of View - Reports effect of angle of view on how well SAFR will recognize a face
- Recognition Quality - Reports effect of distance on how well SAFR will recognize a face
- Learning Quality - Reports effect of distance on how well SAFR will auto-enroll a face from live camera view
The colored bars below the graphic report the quality in terms of:
- Red (low
- Yellow (good)
- Green (best)
The color coding is determined by quality thresholds as follows:
More instructions are provided in the tool itself.