Introduction

If the ML (Machine Learning) Classification feature has been enabled in the Track Engine configuration, then the ML Classification system will be used instead of the traditional weighted scoring system.

ML Classification studies track data to build track models which make predictions about future tracks. This can be aided by the user identifying good, example tracks with known classification types that can be updated via the Add Truth Data feature. We advise following a balanced and economical approach with the number of updated tracks used; too many of one classification type have the potential to skew the models to favour that classification type.

Contents

Enabling ML Classification

To enable ML Classification, navigate to the Topology page and check the setting for the Track Engine. If it is not enabled the Use ML Classification check box will not be selected as in the image below:

Open

Click Edit and select the checkbox and then click Save:

Open

When the ML Classification is enabled, it will allow you to Add Truth Data to tracks on the Track Details screen.

Please refer to AdvanceGuard® Track Queries | Add Truth Data and ClearWay™ Track Queries | Add Truth Data for more information about the Add Truth Data panel.

ML Classification Truth Data

This is a list of the ML Classification data within the system. There will be entries that have been added by default such as those below:

When you Add Truth Data the entry will be added to the bottom of the list.

You are able to delete an entry:

1. Click the Edit button:
   
   

    

2. Click on the entry that you wish to remove and click the bin icon:
   
   

    

3. Then click the Save button:
   
   

    

4. The entry will be permanently removed from the list.

ML Reclassification Rules

ML Reclassification Rules are configured to reclassify a track if its behaviour has changed. For example, a slow-moving track has been classified as a person, but if it starts moving quickly this would cause the track to be re-classified. Most likely the new classification would be a vehicle, the type of which would depend upon the weight of the object.

You can add more rules:

1. Click the Edit button:
   
   

    

2. Click the icon:
   
   

    

3. An Add ML Reclassification Rule dialog will open:
   
   

    

4. here you can add a classification type:
   
   

    

5. Select an attribute:
   
   

    

6. Give that attribute a condition:
   
   

    

7. Add a value to that condition:
   
   

    

8. Click OK and the Save to add that ML Classification Rule to the list:
   
   

    

9. You can remove an ML Reclassification Rule by highlighting an entry and clicking the bin icon whilst in Edit mode.

10. You can also reset the ML Reclassification Rules to their defaults by clicking the reset icon:
    
    

     

11. Clicking Save will return the list of rules to their defaults:
    
    

Related Information