• State Resolved
  • Locked Locked
  • Replies 3 replies
  • Subscribers 31 subscribers
  • Views 426 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

IWR1642BOOST: IWR1642 comparison with other chips like AWR1642 or IWR6843 for people counting application

amin merati
Intellectual 470 points
Part Number: IWR1642BOOST
Other Parts Discussed in Thread: AWR1642, IWR6843

Hi there,

I have got one IWR1642Boost and have been playing with it for the purpose of people counting. 

I have been trying to optimize the accuracy of people counting by some parameters in config file. It seems very hard to get optimized performance to get very high accurate people count specially when people are walking side by side with gap of 5-10 cm between them and they appear in the range of radar and seen as one big lump and tracker sees them as one person all the way through.

Optimizing the parameters turns out to be difficult and it sometime gives people count of 3 instead of 2 and some other time it gives 1 instead of 2.

I know that this is somehow to do with angular resolution which depends on number of transmitter and receiver antennas in MIMO radar.

I also wonder about other mmWave chips such AWR1642? or IWR6843 which is used in lab 19 for people counting?

I looked at block diagram and the only thing I found was that some of mmWave chips are using hardware accelerator instead of DSP, how does this affect?

Could I get better performance or accuarcy with some other of chips? Would I be able to run heavier algorithm on hardware accelerator or DSP?

Please help if I can find the right chip set. We are very interested to get this one working for accurate people count tracking application inside our security gates.

We are one of the biggest global access control company manufacturing security gates for airport, metro, etc. You can see our products in ()

If we manage to get an accurate count specially for our airport gates such as ImmSec shown in this link () 

 We would be using this in huge quantity.

Please help me in this regard.

Amin

 

  • 0
    TI__Mastermind 28055 points

    Hi Amin,

    In the people counting TI design, we set our expected accuracy numbers to:

    +-0 if less than 5 people in the scene

    +-1 if 6-7 people in the scene.

    We expect to hit those numbers 90% of the time, so even for the less than 5 people case, occasionally you will have an issue with incorrect count.

    Some issues that can cause the count to become incorrect:

    • The allocation thresholds are too low, and extra tracks (ghosts) are created as a result - in this case, increase the thresholds as described in the customization guide
    • People are close together, and appear as one target - Decrease the gating parameters, this will limit the size of single track, essentially limiting the number of points that can be associated with a track.  Then there will be points left over to allocate a new track - in general, we expect density to be max 3 people per square meter - see TID for more details

    Regards,

    Justin

  • 0 in reply to Justin Curewitz
    TI__Mastermind 28055 points
    Hi Amin,

    Is your question resolved?

    Regards,
    Justin
  • 0 in reply to Justin Curewitz
    Intellectual 470 points
    Hi Justin,

    Could you please tell me what is the difference between IWR series and AWR? Is one series better than the other?
    Thanks
    Amin