SAFR is able to count how many persons have passed through one or more entrances and add the sum of inbound and outbound traffic to compute current occupancy. SAFR does this by using one or more analytics combined:
- Direction of Travel
- Person Detection
- Face Detection
- Face Recognition
Direction of travel is the key analytic. Without it occupancy counting does not work. There are other ways to implement person counting but SAFR's occupancy reports rely upon this analytic.
You can use either Face or Person detection to count. When using face detection, counting requires 2 cameras to count both inbound and outbound traffic. When using Person detection, only a single camera is needed to bound both inbound and outbound traffic.
Face recognition can optionally be applied to perform watchlist notifications and related functions.
This article explains how to setup occupancy counting with the specific goal of tuning the settings to obtain desired accuracy. When configured correctly it is not unusual to get accuracies of 99,7% for traffic flow. That is, of the total number of subjects passing through, SAFR is off by only 0.3% to the actual count of inbound and outbound (mileage may vary).
The article focuses on using an input file to configure occupancy in order to allow reputable process to test and tune. This is the recommended procedure but only required one representative sample of each unique situation in your site.
This article can also be used to see how to set up occupancy counting in general but does not go into depth on the configuration and setup of reports.
Configuring Occupancy Counting
This article is explained by means of the following video tutorials. You may also choose to reference the following prerequisites:
- Process Video Files In Background - Part1 - Config Occupancy
- Process Video Files In Background - Part2 - Push To Virgo And Change To Type File
- Process Video Files In Background - Part2 - Person Counting Close