To configure Neuralcounter, do the following:

1. If a camera supports multistreaming, select the stream to apply the detection tool to (1). 
   
2. If you need to outline objects in the preview window, select Yes in the Detected Objects parameter (2).
3. Set the recognition threshold for objects in percent (3). If the recognition probability falls below the specified value, the data will be ignored. The higher the value, the higher the accuracy — for the cost of sensitivity.

4. Set the interval between the analyzed frames in seconds (4). The value should be within the range of 0,05 – 30.

5. Set the minimum number of frames with excessive numbers of objects for Neuralcounter to trigger (8). The value should be within the range of 2 – 20.

   The default values (3 frames and 1 second) indicate that Neuralcounter will analyze one frame every second. If Neuralcounter detects more objects than the specified threshold value on 3 frames, then it triggers.

   
   

6. Select the processor for the neural network - CPU, one of GPUs, or Intel NCS (5, see Hardware requirements for neural analytics operation). 

7. Select the neural network file (6).

   For correct neural network operation under Linux, place the corresponding file in the /opt/AxxonSoft/AxxonNext/ directory.

   
   

8. Set the triggering condition for the neural counter:

   1. In the Number of alarm objects field, set the threshold value for the number of objects in FoV (7). 

   2. In the Trigger upon count field, select the condition polarity: whether triggering should occur on exceeding the threshold, or dropping below it (9). 

9. In the preview window, you can set the detection zones with the help of anchor points much like privacy masks in Scene Analytics (see Setting General Zones for Scene Analytics). By default, the entire FoV is a detection zone.
   
10. Click Apply.