To configure License plate recognition (RR), do as follows:

1. Download Addon RR LPR from the website and install it.

   To make the detection tool work under Windows Server 2012 R2, make sure to install the Media Foundation component.

   
   

2. Create the detection object you need and select it (1). 
   

3. If you need to use this detection tool for real-time license plate recognition, set the corresponding parameter to Yes (2) (see Configuring real-time vehicle license plate recognition).

4. To record metadata to the database, select Yes from the Record Objects tracking list (3).
5. If a camera supports multi-streaming, select the stream for which detection is needed. Selecting a low-quality video stream reduces the load on the Server (4).
   

6. Select the Algorithm for detecting vehicle direction (5):

   1. By LP coordinates: if LP coordinates change position from top to bottom, the vehicle moves to the camera. If LP coordinates change position from bottom to top, the vehicle moves from the camera.
   2. By LP scale change: if LP scale increases, the vehicle moves to the camera. If LP scale decreases, the vehicle moves from the camera. 
7. In the corresponding fields, select from the list one or more countries whose license plates you need to recognize (6).
8. Select the processor for decoding video streams (7). When you select a GPU, a discrete graphics card takes priority (when decoding with NVIDIA NVDEC chips). If there is no appropriate GPU, the decoding will be done with Intel Quick Sync Video technology. Otherwise, the CPU resources will be used for decoding.
9. Set the number of frames processed per second by the detector (8). The value should be in the range [0.016, 100].
10. Set the minimum quality of ANPR (9). The higher the minimum recognition quality is, the fewer false detections will occur.

11. Select the processor for the detection tool: the CPU or one of NVIDIA GPUs (10) (see General Information on Configuring Detection).
    

    It may take up to one minute to launch the algorithm on NVIDIA GPU.

    
    

12. In the Force report timeout field (11) specify the time interval between the initial recognition and event registration. The 0 value sets the event registration to the moment when the track disappears from FOV.
    

    Skip this step with License plate recognition – Search in archive (RR).

    
    

13. By default, the analyzed frames are compressed to a specified size (12, 1920 pixels on the longer side). This is how it works:

    1. If the longer side of the source image exceeds the value specified in the Frame size change field, it is divided by two.

    2. If the resulting resolution falls below the specified value, this resolution will be used further.

    3. If the resulting resolution still exceeds the specified value, it is divided by two, etc.

       For example, the source image resolution is 2048*1536, and the specified value is set to 1000. In this case, the source resolution will be halved two times (down to 512*384), as after the first division, the number of pixels on the longer side exceeds the limit (1024 > 1000).

       
       

       If detection is performed on a higher resolution stream and detection errors occur, it is recommended to reduce the compression.

       

14. Specify the maximum number of recognition threads (13). If the value is 0, the recognition process will occur in the same thread that starts it.

    The cumulative value of this parameters across all LPR detection tools should not exceed the number of CPU cores and is limited to 100.

    
    

15. You can configure an ANPR zone in FOV. The zone is resized by moving the anchor points . 
    

    For your convenience, you can click the button and configure the mask on a still frame. To undo, click this button again.

    
    

    Detection zone is displayed by default. You can click the button to hide the zone. To undo, click this button again.

    
    

16. Click the Apply button.

Configuration of License plate recognition (RR) is now complete.