• State Resolved
  • Locked Locked
  • Replies 5 replies
  • Subscribers 29 subscribers
  • Views 457 views
  • Users 0 members are here
Support feedback
Options
Options
Related

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.

Original question:

IWR1843BOOST: Medium Range Radar: Why there is mismatch in tracked object length in GUI and log generated?

IWR1843BOOST: Medium Range Radar: Wrong xSize and ySize received

Asha Jose
Prodigy 146 points
Part Number: IWR1843BOOST

We have captured a data of a person walking in front of the sensor up and down. The output of mrr parser is attached below.

parsed_txt.txt

Similary we have also run the mrr gui application for the same scenario and we have obtained the following output 

On analyzing both the data xSize and ySize is found to be around >1.5m most of the time. For most of the frames xSize and ySize is found to be exactly the same

for example , please find below a sample from the mrr output file attached along with this thread


"data": [
              {
                "x": 24,
                "y": 2344,
                "xd": -1,
                "yd": 0,
                "xSize": 261,
                "ySize": 261
              },
              {
                "x": 10098,
                "y": 5701,
                "xd": 94,
                "yd": -166,
                "xSize": 261,
                "ySize": 261
              },
              {
                "x": -19,
                "y": 2956,
                "xd": -4,
                "yd": 0,
                "xSize": 261,
                "ySize": 261
              },
              {
                "x": 334,
                "y": 3639,
                "xd": -7,
                "yd": 0,
                "xSize": 234,
                "ySize": 244
              },
              {
                "x": -203,
                "y": 1274,
                "xd": 0,
                "yd": 0,
                "xSize": 261,
                "ySize": 261
              }
            ]

The only moving object in the scenario is a person which cannot have width >1.5 m . 

We are getting wrong xSize and ySize. From the output of mrr parser we are not at all able to find the width and length of the person walking in the scenario. What is the reason for obtaining this wrong information? 

  • 0
    TI__Guru*** 149154 points

    Hi,

    If the goal is to determine the width and length of the person for classification purpose than this is not the right demo to use.

    The basic tracker implemented is not robust enough for this use case.

    Thank you

    Cesar

  • 0 in reply to Cesar
    Prodigy 146 points

    Hi ,

    The goal  is to determine the width length of vehicles like motor cycle, car, SUV , and truck.  The reason behind man walking is very much close to bike. We also need the tracking info the vehicle to determine is the distance.  Which demo would you be recommending to use? 

    We have tested with car then also the same issue was persistent .  Kindly suggest what shall be done.

  • +1 in reply to Asha Jose
    TI__Guru*** 149154 points

    Hi,

    This demo will not provide you this type of classification: width, length of vehicle.

    This demo will provide you range, velocity information + tracking.

    We don't have any demo that provides vehicle length, width information

    You would probably need to develop your own custom algorithms.

    Unfortunately our support team does not have additional information

    thank you

    Cesar

  • 0 in reply to Cesar
    Prodigy 146 points

    Hi,

    So may I ask what does cluster size indicate in the scene in Medium Range radar type 3 tlv? 

    According to my understanding,

    • Medium Range Radar demo shows the cluster size of the tracked object. ie vehicle/object in type 3 tlv packet.
    • so I assume that xSize is width of the object and
    • ySize is length / height of the object in the scene. 

    Please correct me if I'm wrong. 
    Also, if Im wrong please tell me what is the cluster Size indicate in the scene?

  • +1 in reply to Asha Jose
    TI__Guru*** 149154 points

    Hi

    DBscan Clustering is the algorithm use.

    There are a lot of resources on the web explaining this algorithm. For example

    https://www.youtube.com/watch?v=RDZUdRSDOok

    It is important to understand that the size of the cluster depends on the number of detected objects. You can't extrapolate this to mean size of vehicle. For example, you could have only 2 detected points from a vehicle. So the cluster size will be small and not reflect the size of the vehicle

    thank you

    Cesar