State estimation of detected objects with drivenet


Driveworks object detection module is missing further details about detected object state: position, speed, acceleration etc. dwObjectCamera structure has obstacle structure that contains those fields but they appear always to be 0.
Is this expected and state estimation is out of scope of drivenet? Or is there some undocumented setup to get the following information?

It is required to know whether driveworks has something that covers/generates said data or does it need custom implementation to fill that in.

Hi marek.piirikivi,

How do you check obstacle? Which sample are you using? Thanks!

I am using drivenet (looked for both drivenet samples: with single camera and with n-cameras). All I can get is classification and bounding box coordinates. From there, using object detection module I can get detections that contain obstacle data.

Did you mean this topic was clarified?

No, it is fussy whether out-of-the box there exists a module/method that fills in this obstacle info. It would be beneficial to know each detected object position, velocity, acceleration and direction of movement. Should I expect that from drivenet/object detection or do I have to implement it on my own?

The obstacle info in dwObjectCamera will be available in the next release as I mentioned in below topic. Thanks!