IWR1642: One target is identified as multiple

Hello.

I would look to modify some of the tracking parameters.  It is possible you may have set the gating parameters small enough such that it allocates one vehicle track for an object that might be smaller than one.  You can refer to the other tracking commands in this document here.

Sincerely,

Santosh

Thank you for your reply.I will try to increase the gating parameters.

However, if the gating parameters are increased, there will be a problem that two vehicles in some adjacent lanes may be identified as one vehicle. In this case, is it a priority to ensure that vehicles running in separate lanes are not identified as two? Then adjust other parameters so that two vehicles in the adjacent lane can be distinguished?

Regards,

ctc

It will take tuning the gating parameters such that it knows that a certain max size and velocity should get a track.  But if your priority is being able to track two vehicles in adjacent lanes without the tracker confusing the tracks, I would look to this command tune its parameters as close as possible to the size of the vehicle.

Hope this helps,

Santosh

Hi,Santosh.

My gating parameter settings refer to the default settings below

[1]The default is the sequential configuration of 16, 12, 8 and 0. This configuration is rarely distinguished into two targets when a single lane vehicle travels. However, when vehicles in multiple lanes move together, it is difficult to distinguish vehicles in adjacent lanes. Where half of the LengthLimit value represents the expected maximum vehicle length? For example, the default value is 12, indicating that the maximum vehicle length is 6 meters. Similarly, WidthLimit is 8, indicating that the maximum vehicle width is 4 meters?

[2]If I want to give priority to ensuring the detection effect of adjacent lanes. So how can I avoid a single lane vehicle being detected as two targets? What other parameters can I modify?

Regards,

ctc

Hello.

This is a much older version of the lab and so I would have to look into this further and I can get back to you early next week.  But if your priority is detecting unique vehicles in adjacent lanes, I would tune the gating parameters first and then adjust the allocation parameter to be more scrutinizing when allocating a track.  The allocation parameters allocate tracks based on SNR and point cloud so maybe adjusting this as well to be more scrutinizing can help.

Sincerely,

Santosh

Hello.

ctc ctc said:

Where half of the LengthLimit value represents the expected maximum vehicle length? For example, the default value is 12, indicating that the maximum vehicle length is 6 meters. Similarly, WidthLimit is 8, indicating that the maximum vehicle width is 4 meters?

This should not be the case.  In the image you provided it says that the ellipsoid will not stretch past 8m if the width limit is set to 8, indicating that the widthLimit value is the width of the object, but the width of the track will stretch 4m from the center of the vehicle.  If you look at the tracker layer guide in the 3D People Counting Lab in the toolbox, you can find more information on what the individual parameters for each configuration command represent.  You can modify these individual parameters based on the performance you are seeing. 

ctc ctc said:

If I want to give priority to ensuring the detection effect of adjacent lanes. So how can I avoid a single lane vehicle being detected as two targets? What other parameters can I modify?

As I said in my previous response, I would try to increase the parameters for the gating configuration command to account for this.  If you increase the width(as the vehicles are next to each other horizontally) too much, you may end up seeing 1 track when there are two vehicles.  So you will have to experimentally determine what value is correct for your use case.

Hope this helps,

Santosh