Bounding Box Metadata Jumps Based on Batch Size

Please provide complete information as applicable to your setup.

• Hardware Platform (Jetson / GPU)
• DeepStream Version
• JetPack Version (valid for Jetson only)
• TensorRT Version
• NVIDIA GPU Driver Version (valid for GPU only)
• Issue Type( questions, new requirements, bugs)
• How to reproduce the issue ? (This is for bugs. Including which sample app is using, the configuration files content, the command line used and other details for reproducing)

  • Use the DeepStream reference application.
  • Configure a pipeline to use two sources and also makes use of the MSG_CONV_BROKER sink, being sure to set a debug directory for the resulting converted messages. Also configure the sink to use the DEEPSTREAM_MINIMAL schema. Also make use of a primary inference engine and a tracker (in my case, the tracker is NvDCF and makes use of the libnvds_nvmultiobjectracker lib, with batch processing enabled along with past frame).
  • Note that, at seemingly random intervals, a tracked object’s bounding box seems to have its (x, y) coordinates (i.e. “left” and “top”) doubled (unless the proper value is actually the doubled one in which case it’s halved).

I’ve checked and confirmed that the “scaleW” and “scaleH” values used in creating the DEEPSTREAM_MINIMAL payloads are both 1.0.

I don’t have this issue when using only a single source, so given that the difference is a factor of 2, I’m inclined to believe that something about the number of sources is doing something funny.

Perhaps I’ve configured something incorrectly, but I can’t figure out what since the bounding box jumps at random.

Any help would be greatly appreciated.

Thank you.

Can you have a try with latest DS 6.0.1? Can you reproduce it in DS samples?

I tried this with DS 6.0.1 and it seems to be gone. That said, I can’t be 100% sure whether it was the update that fixed it or whether I just had some bad config somewhere.

I built the app up from “deepstream-app” after the 6.0.1 install piece-by-piece until I returned to the “original” state by copying in code and configs, but it’s possible I might have missed something.

Either way, it was a worthwhile exercise and the issue seems to be gone. Hopefully it was the upgrade that fixed it and this should be the first port of call for anyone who might also experience this same issue.

Thank you for your help.

Great! Glad to know the upgrade fixed the issue.

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.