Introduction

Vectors have the capability to perform at an extremely high accuracy level, and it is common for devices to track and count people perfectly for very long periods of time before they eventually miss someone or make a double count mistake. For this reason we quote high expected accuracy levels of up to 99.5%.

But it is vital that devices are installed and configured correctly in order to achieve those levels of accuracy.

The built in validation functionality allows you to playback a recorded file in order to manually count and compare with the Vectors automated counts in order to judge the accuracy percentage.

The ability to be able to watch how a device actually performed along with the ability to slow down and speed up playback as required makes it very easy to determine an accuracy level, and importantly, being able to see where a device went wrong, makes it easy to alter the settings in order to improve that accuracy level.

If a device is installed well, and configured optimally, the accuracy should be revealed to be at a very high level from day one, but where accuracy is not acceptable the recording should point to the settings changes required.

It is important to state that by watching the video, a user will be able to see where a Vector is falling down in accurate counting and make meaningful settings changes. Partners are expected to configure their own devices as part of their full service offering to their customers - this includes finessing the setup as required. Please do not raise frivolous support tickets for poor accuracy without first making the effort to understand and correct the issue yourself.

 

Pre-requisites

Before going through the process of validating a device it is important to spend some time ensuring that it is installed and configured optimally. Validating a device that you know is installed or configured less that optimally will simply reveal an accuracy that you already know will not be perfect, and your time will be better put to use going through the setup of that devices first, also noting any red flags with regards to the placement of that device relative to the door.

Once you have configured a device correctly, a validation session will then tell you the accuracy level and will enable you to make the decision to amend and improve the setup as necessary in order to improve accuracy further.

In order to validate you must first have a validation file available to view. This means you need to schedule a recording over a suitable time frame. 

If using Estate Manager to validate, the recording needs to be downloaded to Estate Manager first.
If using the RIFT exe software, the recording needs to be downloaded to your local hard drive first.

 

Notes

It is not possible to validate node device recordings when connected via the web browser version of RIFT. To validate node recordings use either an Estate Manager connection, or use the RIFT exe standalone software.

 

Validation.

To Validate a recording click the Validation tab and then highlight the relevant file, and click 'Open':

 

It will take a few seconds to open the file, so be patient:

 

The first step in the Validation process is the preview window:

In the Recording Preview window you can play the video and go back and forth along the timeline in order to confirm that the recording is suitable for a Validation. Just use the timeline buttons at the bottom of the window to navigate around.

You can also switch between video views if the validation contains views from more than one device using the buttons on the 'Devices' window in the top left.

Each of the short black lines on the timeline next to the register name are increments on that register. In the above example, you can see only about 18 count increments in total over the 30 minute recording, so this could be considered a very quiet scene and would not provide a statistically relevant result.

If the file is deemed to be unsuitable click the 'Back' button. But if it is ok to proceed, click the 'New' button to start a New validation session.

 

The Validation Session Creation step is next:

At this point you should select the registers that you want to validate against - simply place a tick in the boxes on the right. In most cases.

There will probably be just two count line registers in the most cases - for counting IN and OUT - but for more complicated applications you can choose to validate up to six registers as a time.

When ready to begin, click the 'Next' button.

 

Before you can start the actual validation you can choose to enable or disable certain options, and lock them so that they cannot be changed during the validation session.

For the majority of customers, everything here should be left on default enabled and unlocked, but this step provides an extra level of functionality for those that need it. If you are worried about people being influenced by some of the things being seen then they can be disabled.

These options are mirrored from within the validation session but choosing to disable and lock them means they cannot be re-enabled later.

 

Show Visuals - This shows/hides any lines and zones that are configured. This is provided so that auditors can watch people going through a doorway or other opening without being influenced by simply watching people cross the lines or enter the zones. This ensures a good overall accuracy but it should be pointed out that hiding the visuals will not allow the auditor to see how any mistakes were made - i.e. when a line was missed, or a zone was not entered.

Show Targets - This shows/hides the targets associated with the tracked people. The idea with this is that auditors do not simply watch targets and ensures that the video view of the actual people walking is being validated.

Show Targets Height - This shows/hides the additional height information alongside each target. Showing this will be required if attempting to validate adult/child counting.

Show Targets Paths - This shows/hides the track in front of the person and enables auditors to see what is going to happen before it happens. 

Show Targets Tails - This shows/hides the track behind a person and enables auditors to see where a person has been tracked previously.

Show Targets Over Video - This only applies to wide openings of two or more devices where the video from the master may not see the extreme left or right of the combined field of view from all units. Where a person is not viewable you can choose to not show the corresponding targets being generated from the time of flight data. This should be used with care as any line crossings or other register changes will still occur. Disabling this is not recommended. It should also be noted that Vector validations can now contain more than one video feed so with careful scheduling of recordings to ensure all the available field of view is covered by video views, this option will not be required.

Show System Counts - This shows/hides the main register increments provided by the Vector and allows 

Show Field of View - This shows/hides the time of flight field of view. It should be pointed out that the video field of view is larger than the active time of flight field of view so care should be taken when hiding the is you will not be able to see if people are being missed because they are not trackable within the active field of view.

 

When you're ready to start validating click the 'Goto Validation' button.

 

You're now ready to start Validating.

Press the play button and watch the people walking around. For regular IN and OUT counting, as you see people enter or exit, simply record a corresponding manual IN or OUT count by clicking the relevant count increment buttons or pressing the relevant key on your keyboard. In the example above you can see that the 'Q' key adds a manual count to the green line and the 'W' key adds a manual count to the red line.

As the recording plays the yellow timeline indicator will show where in the redoing you are, and you can use the speed controls to speed up and slow down the recording as appropriate. 

Once the recording is finished you will have a direct comparison between the System counts and your own manually added counts. Assuming your validating was correct your own User counts will indicate what the Vector(s) should have counted and the system counts will be what the Vector(s) actually counted. The closer the values the better.

If the recording is valid, i.e. it contains enough counts to be statistically relevant and the accuracy is good, you can then click the 'Complete' button. But if the recording is quiet with very few people, or the accuracy as currently configured is poor, click the 'Reject' button.

Importantly, where a mistake is made, the video view shows what happened and you should be able to work out why it happened. Just look at the positioning of the counting lines and the target trajectory as a first step and if the target is crossing the lines correctly but is not being counted, ensure that the count lines underlying settings are appropriate for that target - i.e. height, staff, count mode.