To configure the Object presence detector module:

1. Go to the settings panel of the Object presence detector object.
   
2. To receive events about the end of object presence only after the time period from the Alarm duration, s setting, set the Track alarm end checkbox. By default, the checkbox is clear, which means that events about the end of object presence are received immediately.
3. In the Alarm duration, s field, enter the time period in seconds during which the repeated object presence alarm events are ignored. By default, the setting isn't available for editing. To enable it, you must set the Track alarm end checkbox (see the previous step).
   

   If you set the Track alarm end checkbox, the events about the end of object presence are received after the time specified in the Alarm duration, s setting. If, until the end of this period, the current object presence is completed and the new one is detected, the detector doesn't generate events about completing the current and starting the new alarm. The new object presence event is received after the current alarm is completed.

   
   
4. Set the Generate alarm on camera checkbox if you want to initiate an alarm on the camera when the detector generates an event.
5. Click the button to the right of the Tracking model field, and in the standard Windows Explorer window that opens, select the file of the trained neural network (for information on how to train neural networks, see Starting training and retrieving results).
   

   The service trains neural networks to work with specific camera angles and objects. It's important to note that a neural network trained on one scene is not intended for use in another, even if they appear visually similar. In such cases, the accuracy and performance of the neural network are not guaranteed.

   
   

6. In the Detection sensitivity [0, 100] field, enter the detector sensitivity—an integer value in the range from 0 to 100. This parameter determines the level of reliability at which an alarm is generated when object presence is detected: the higher the sensitivity, the higher the recognition threshold. For example, at sensitivity 100, the neural network of the detector must get 100% certainty for some period of time during the frame processing. The value of the Detection sensitivity field is determined experimentally.

7. In the Frames processed per second [0.016, 100] field, set the number of frames per second that the neural network processes in the range of 0.016–100. This value affects the processor load: the higher the value, the greater the load. It also affects the operation speed of the algorithm: the higher the frame rate, the faster the required number of frames are collected for making a decision and generating an alarm, if required.

8. In the Number of frames for analysis and output [5, 20] field, enter the minimum number of frames in the range from 5 to 20 for the detector to analyze before generating an alarm. The higher the value, the more reliable the result of the detector operation. At the same time, if the value is too high, the short-term presence of an object can be missed. If you enter a value less than the minimum, the value is set to 5.

9. Set the Scanning mode checkbox to detect small objects. If you enable this mode, the load on the system increases. That is why we recommend setting the small number of frames processed per second in step 7. By default, the checkbox is clear. For more information on the scanning mode, see Configure the scanning mode.
   

   You cannot configure the scanning windows for the Object presence detector module. The default parameters are used. When you use the scanning mode, we recommend using a full frame without specifying the area of interest in step 11.

   
   
10. From the Device drop-down list, select the device on which the neural network will operate: the CPU, one of the NVIDIA GPUs, or one of the Intel GPUs. Auto (the default value)—the device is selected automatically: The NVIDIA GPU gets the highest priority, followed by the Intel GPU, then the CPU.
    

    We recommend using the GPU. It can take several minutes to launch the algorithm on the NVIDIA GPU after you apply the settings. You can use caching to speed up future launches (see Configuring the speedup of neural analytics launch on GPU). In the Detector Pack subsystem of version 2.0, the support of Intel HDDL is removed, therefore, when you update from version 1.0, the Not supported option is selected automatically instead of this device variant, and detectors don't operate. To resume the detector operation, select the required device from the list.

    
    
11. Specify the area of interest on the video image:
    1. Click the Stop video button to capture the frame of the video image.
       
       
    2. Click the Area of interest button.

    3. On the captured video frame, sequentially set the anchor points of the area the situation in which you want to analyze using the left mouse button. After you add the area, the remaining part of the video image is darkened. If you don't specify the area of interest, the entire frame is analyzed.

    4. Click the OK button.
       

       To remove the area, click the button next to the Area of interest button. You can add only one area. If you try to add the second area, the first area is deleted.

       
       
    5. Click the Apply button.

Configuration of the Object presence detector module is complete.