Versions Compared

Old Version 3

changes.mady.by.user Evgeniya Kononova

Saved on

compared with

New Version 4

changes.mady.by.user Bahdan Tsiulkou

Saved on

Key

  • This line was added.
  • This line was removed.
  • Formatting was changed.
Section


Column
width50%


Panel
borderColor#CCCCCC
bgColor#FFFFFF
titleBGColor#F0F0F0
borderStylesolid
titleOn the page:
Table of Contents



Column


The Neurotracker module program module registers object tracks in the camera FOV during recording using a the neural network and saves them to the VMDA metadata storage (see Creating and configuring VMDA metadata storage).
Configuration The configuration of the Neurotracker program module includes: main and additional settings of the detection tooldetector, selection of the area the area of interest, configuration of the neurofilter configuration. 

You can configure the Neurotracker program module on the settings panel of the the Neurotracker object that is created on the basis of the Camera object on the Hardware tab of the System settings dialog window.

Main settings of the

...

detector

You can configure the main settings of the detection tool detector on the Main settings tab on the settings panel of the Neurotracker object.

  1. Set the Generate event on appearance/disappearance of the track checkbox to generate an event when an object (track) appears in the frame and disappears from the frame.

    Info
    titleNote

    The track appearance/disappearance events are generated only in the debug window (see Start the debug window). They

    2. are not

  2. aren't displayed in the Event viewer.

  3. Set the Show objects on image checkbox to highlight the detected object with a frame when viewing live video.

  4. Set the Save tracks to show in archive checkbox to highlight the detected object with a frame when viewing the archive.

    Info
    titleNote

    This parameter

    5. does not

  5. doesn't affect the VMDA search and is used just for the visualization. For this parameter,

    6. the 

  6. the titles database is used.

  7. Set the Model quantization checkbox to enable model quantization. By default, the checkbox is
    8. clear
  8. cleared. This parameter allows you to reduce the consumption of the GPU processing power.
    Info
    titleNote
    1. AxxonSoft conducted a study in which a neural network model was trained to identify the characteristics of the detected object. The following results of the study were obtained: model quantization can lead to both an increase in the percentage of recognition and a decrease. This is due to the generalization of the mathematical model. The difference in detection ranges within ±1.5%, and the difference in object identification ranges within ±2%.
    2. Model quantization is only applicable for NVIDIA GPUs.
    3. The first launch of
    9. a detection tool
    1. the detector with quantization enabled
    10. may
    1. can take longer than
    11. a
    1. the standard launch.
    2. If GPU caching is used, the next time
    12. a detection tool
    1. the detector with quantization will run without delay.
  13. From the Object type drop-down list, select the object type for analysis:
    • Human
    14. —the camera
    • —camera is
    15. directed
    • pointed at
    16. a
    • the person at the angle of 100-160°;
    • Human (top-down view)
    17. —the camera is directed at a
    • —camera is pointed at the person from above at a slight angle;
    • People view from above (Nano)—camera is pointed at the person from above at a slight angle, small network size;
    • People view from above (Medium)—camera is pointed at the person from above at a slight angle, average network size;
    • People view from above (Large)—camera is pointed at the person from above at a slight angle, large network size;
    • Vehicle
    18. —the camera
    • —camera is
    19. directed
    • pointed at
    20. a
    • the vehicle at the angle of 100-160°;
    • Person and vehicle (Nano)
    21. —person
    • —detects person and vehicle
    22. recognition
    • , small
    23. neural
    • network size;
    • Person and vehicle (Medium)
    24. —person
    • —detects person and vehicle
    25. recognition
    • ,
    26. medium neural
    • average network size;
    • Person and vehicle (Large)
    27. —person
    • —detects person and vehicle
    28. recognition
    • , large
    29. neural
    • network size.
      Info
      titleNote

      Neural networks are named taking into account the objects they detect. The names can include the size of the neural network (Nano, Medium, Large), which indicates the amount of consumed resources. The larger the neural network, the higher the accuracy of the object recognition but the greater the load on the CPU.

  30. By default, the standard
    31. (default)
  31. neural network is initialized according to the selected object
    32. selected in the Object type drop-down list and the device selected in the Device drop-down list. The standard neural networks for different processor types are selected automatically. If you use a custom neural network, click the Image Removed
  32. type on step 5 and device on step 7. You must not select manually standard networks for different processor types since it is performed automatically. If you have the unique neural network for use, click the Image Added button to the right of the Tracking model field
    33.  
  33. and specify its file in the standard Windows Explorer window
    34. , specify the path to the file
  34. that opens.
    Note
    titleAttention!

    To train

    35. a

  35. the neural network, contact

    36. the 

  36. AxxonSoft technical support

    37.  

  37. (

    38. see

  38. see Data collection requirements for neural network training).

    39. A

  39. The use of the trained neural network

    40. trained

  40. for a

    41. specific

  41. particular scene allows you to detect only objects of a certain type

    42. only

  42. (for example, a person, a cyclist, a motorcyclist, and so on).

  43. From
    44. the 
  44. the Device drop-down list, select the
    45. device
  45. one on which
    46. the neural
  46. the neural network
    47.  
  47. will operate: the CPU, one of the NVIDIA GPUs, or one of the Intel GPUs. Auto (default value)
    48. —the device
  48. —device is selected automatically:
    49.  
  49. The NVIDIA
    50.  
  50. GPU
    51. gets
  51. takes the highest priority,
    52. followed by
  52. then goes the Intel GPU, and then the CPU.
    Note
    titleAttention!
    1. We recommend using the GPU.
    2. It
    53. may
    1. 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 Optimizing the operation of neural analytics on GPU).
    2. In the Detector Pack 2.0subsystem, the Intel HDDL support is removed. Thus, when you update from the 1.0 version, the Not supported option is automatically selected instead of this device option, and detectors won't operate. To resume detector operation, select the required device from the list.
  54. From the Process drop-down list, select which objects must be processed by the neural network:
    • All objects—moving and stationary objects;
    • Only moving objects—an object is considered to be moving if, during the entire lifetime of its track, it
    55. has
    • shifted by more than 10% of its width or height.
    56. Using
    • If you use this parameter, you can reduce the number of false positives;
    • Only stationary objects—an object is considered stationary if, during the entire lifetime of its track, it
    57. has
    • shifted by no more than 10% of its width or height. If
    58. a
    • the stationary object starts moving, the
    59. detection tool triggers
    • detector generates an event, and the object is no longer considered stationary.
      Info
      titleNote

      The selection of only moving objects and only stationary objects isn't mutually exclusive, as some tracks cannot be determined as either moving or stationary. First, the neural network detects all objects, and after that, the detector filters out unnecessary tracks in accordance with the selected value of the Process setting.

  60. From the Camera position drop-down list, select:
    1. Wall—objects are detected only if their lower part gets into the area of interest specified in the
    61. detection tool
    1. detector settings.
    2. Ceiling—objects are detected even if their lower part doesn't get into the area of interest specified in the
    62. detection tool
    1. detector settings.

Selecting the area of interest

  1. Click the Settings button.
    2. The Detection
  2. As a result, the detector settings window opens.
    Image Modified
    3. Click
  3. In the Detection settings window, click the Stop video button (1) to pause the playback and capture the frame of the video image.
  4. Click the Area of interest button (2) to specify the area of interest. The button
    5. will be
  5. is highlighted in blue.
    Image Modified
  6. On the captured frame of the video image, use the mouse to sequentially set the anchor points of the area (1)
    7. ,
  7. in which the objects
    8. will be
  8. are detected. The rest of the frame
    9. will be
  9. is faded.
    10. You
  10. There can
    11. add 
  11. be only one area of interest. To delete an area, click the Image Modified button. If you don't specify the area of interest, the entire frame is analyzed.
  12. Click the OK button (2) to close the Detection settings window and return to the settings panel of the
    13. Neurotracker object
  13. detector.

Additional settings

  1. Go to the Additional settings tab on the settings panel of
    2. the Neurotracker object
  2. the neurotracker.
    Image Modified

  3. In the Recognition threshold [0,100] field, specify the neurocounter sensitivity—an integer

    4. value

  4. number in the range from 0 to 100.

    Info
    titleNote

    The neurotracker sensitivity is determined experimentally. The lower the sensitivity, the higher the probability of false alarms. The higher the sensitivity, the lower the probability of false alarms, however, some useful tracks can be skipped (see Examples of configuring neural tracker for solving typical tasks).

  5. In the Frames processed per second [0.016, 100] field, specify the number of frames processed per second by the neural network in the range from 0.016 to 100. For all other frames the interpolation
    6. will be
  6. is performed—finding intermediate values by the available discrete set of its known values. The greater the value of the parameter, the more accurate the
    7. detection tool operation
  7. tracking, but the higher the load on the processor.
    Info
    titleNote

    The recommended value is at least 6 FPS. For fast moving objects (running person, vehicle)—at least 12 FPS (see Examples of configuring neural tracker for solving typical tasks).

  8. In the Minimum number of triggering [2, 100] field, specify the minimum number of the neurotracker
    9. triggers required
  9. triggerings to display the object track. The higher the value of this parameter, the longer it takes from the object detection moment to the display of its track.
    10. A
  10. The low value of this parameter can lead to false positives. The default value is 6. The value range is from 2
    11. -
  11. to 100. The entered
    12. value
  12. number that is greater than the maximum value or less than the minimum value from the specified range
    13. ,
  13. is automatically adjusted to the maximum or minimum value, respectively.

  14. In the Track hold time (s) field, specify the time in seconds after which the object track is considered lost in the range from 0.3 to 1000. This parameter is useful in situations

    15. where

  15. when one object in the frame temporarily overlaps another. For example, when a large vehicle completely overlaps a small one.

    Info
    titleNote

    If an object (track) is close to the frame boundary, then approximately half of the time specified in

    16. the

  16. the Track hold time (s) field must elapse from the moment the object disappears from the frame until its track is deleted.

  17. Set
    18. the 
  18. the Scanning mode
    19.  
  19. checkbox to detect small objects. If you enable this mode, the load on the system increases.
    20. So
  20. That is why, on step 3, we recommend specifying a small number of frames processed per second
    21. in the Frames processed per second [0.016, 100] field
  21. . By default, the checkbox is
    22. clear
  22. cleared. For more information on the scanning mode, see Configuring the Scanning mode.
  23. If necessary, specify the class of the detected object in the Target classes field. If you want to display tracks of several classes, specify them separated by a comma with a space. For example, 1, 10.
    The numerical values of classes for the embedded neural networks: 1—Human/Human (top view), 10—Vehicle.
    Info
    titleNote
    1. If you leave the field blank, the tracks of all available classes from the neural network
    24. will be
    1. are displayed (Object
    25.  
    1. type, Neural network file).
    2. If you specify a class/classes from the neural network, the tracks of the specified class/classes
    26. will be
    1. are displayed (Object
    27.  
    1. type, Neural network file).
    2. If you specify a class/classes from the neural network and a class/classes missing from the neural network, the tracks of a class/classes from the neural network
    28. will be
    1. are displayed (Object
    29.  
    1. type, Neural network file).

    2. If you specify a class/classes missing from the neural network, the tracks of all available classes from the neural network

    30. will be

    1. are displayed (Object

    31.  

    1. type, Neural network file).

Neurofilter

You can use the neurofilter to sort out some of the tracks. For example, the neurotracker detects all freight trucks, and the neurofilter leaves only those tracks that correspond to trucks with cargo door doors open. To configure a the neurofilter, do the following:

  1. Go to the Neurofilter tab on the settings panel of
    2. the Neurotracker object
  2. the neurotracker.
    Image Modified
  3. Set the Enable filtering checkbox to enable the neurofilter. By default, the checkbox is
    4. clear
  4. cleared.
  5. By default, the standard
    6. (default)
  6. neural network
    7. is initialized
  7. according to
    8. the device selected in the Device drop-down list. The standard neural
  8. the selected device on step 4 is initialized. You must not select manually standard networks for different processor
    9. types are selected
  9. types since it is performed automatically. If you
    10. use a custom
  10. have the unique neural network for use, click
    11. the Image Removed button
  11. the Image Added button to the right of
    12. the 
  12. the Tracking model
    13.  field and
  13.  field and specify its file in the standard Windows Explorer window
    14. , specify the path to the file
  14. that opens.
    Note
    titleAttention!

    To train

    15. a

  15. the neural network,

    16. contact the 

  16. contact  AxxonSoft technical support

    17.  

  17. (

    18. see

  18. see Data collection requirements for neural network training).

    19. A

  19. The use of the trained neural network

    20. trained

  20. for a

    21. specific

  21. particular scene allows you to detect only objects of a certain type

    22. only

  22. (for example, a person, a cyclist, a motorcyclist, and so on).

  23. From the Device drop-down list, select the
    24. device
  24. one on which
    25. the neural
  25. the neural network
    26.  
  26. will operate: the CPU, one of the NVIDIA GPUs, or one of the Intel GPUs.
    27.  Auto (default value)—the device is selected automatically: NVIDIA GPU gets the highest priority, followed by Intel GPU, then CPU.

  27. Info
    28. note
  28. title
    29. Attention!
  29. Note
    1. The device for the neurofilter must match the device specified for the neurotracker in
    30. the Device drop-down
    1. step 7 of the main settings.
    31.  
    2. It
    32. may
    1. can take several minutes to launch the algorithm on the NVIDIA GPU after you apply the settings.

  33. Click the Apply button to save the changes.

    Info
    titleNote

    If necessary, create and configure the

    34. NeuroTracker

  34. Neurotracker VMDA

    35. detection tools

  35. detectors on the basis of the Neurotracker object. The procedure of creating and configuring the

    36. NeuroTracker

  36. Neurotracker VMDA

    37. detection tools

  37. detectors is similar to creating and configuring the VMDA

    38. detection tools

  38. detectors for

    39. a

  39. the regular tracker. The only difference is that

    40. it is necessary to

  40. you must create the

    41. NeuroTracker

  41. Neurotracker VMDA

    42. detection tools

  42. detectors on the basis of

    43. the

  43. the Neurotracker object

    44. ,

  44. and not on the basis of the Tracker object (see Creating and configuring the VMDA detection). Also,

    45. if

  45. when you select the Staying in the area for more than 10 sec detector type, the time the object stays in the zone, after which the NeuroTracker VMDA

    46. detection tools are triggered

  46. detectors generate an event, is configured using the LongInZoneTimeout2 registry key, not the LongInZoneTimeout. The

    47. procedure of configuring the

  47. alarm generation mode is set for any type of VMDA

    48. detection tools is

  48. detector similar to the VMDA

    49. detection tools

  49. detector for

    50. a

  50. the regular tracker using the VMDA.oneAlarmPerTrack registry

    51. key 

  51. key (see Registry keys reference guide).

    Image Modified

Configuration of the Configuring the Neurotracker program module is complete.

Tip

If events are periodically received from several objects, then for convenience, you can create and configure we recommend creating and configuring the neurotracker track counters (see Configuring the neurotracker track counter).