depthai-ros icon indicating copy to clipboard operation
depthai-ros copied to clipboard

Published 20 hours ago • verified publisher icon luxonis

Covariance of spatial detections

Open hwgat opened this issue 3 years ago • 4 comments

Hi there,

I am currently using multiple OAK-d cameras on a mobile robot for pedestrian detection. I have used the Spatial Tiny-yolo example as a template and then publish detections on a ros topic using rosbridge.

The tracking package I am using requires geometry_msgs/PoseWithCovariance messages as an input. Is there a way to calculate the covariance of an OAK-d spatial detection?

Cheers,

hwgat avatar Aug 22 '21 23:08 hwgat

I don't think we have it. Let me cross check it and get back to you.

saching13 avatar Aug 23 '21 01:08 saching13

I am also interested in this!

jonathanleinola avatar Oct 25 '21 15:10 jonathanleinola

Has there been an update on this? I am using the internal Object Tracker and I would also be interested in obtaining the covariance of the Tracklets. Stating the documentation the tracker uses Kalman Filters, therefore this information should already be there

ninoweg avatar May 12 '23 15:05 ninoweg

Hello, This comes from an internal fw from Mx where it doesn't come with an option to enable the covariance of the tracklets.

saching13 avatar May 17 '23 15:05 saching13