Question regarding confidence

Please provide complete information as applicable to your setup.

• Hardware Platform (Jetson / GPU)
Jetson Nano Dev Kit 4G
• 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)
• Requirement details( This is for new requirement. Including the module name-for which plugin or for which sample application, the function description)


I’m having a doubt regarding the confidence info element in NvDsObjectMeta — Deepstream Deepstream Version: 5.0 documentation

while l_obj is not None:
        # Casting to pyds.NvDsObjectMeta
        obj_meta = pyds.NvDsObjectMeta.cast(
    except StopIteration:
    obj_counter[obj_meta.class_id] += 1
        l_obj =
    except StopIteration:

The print always just gives -0.10000000149011612. What is the meaning of this?

And if I’m allowed to ask: How to interpret ntp_timestamp?


Am I right to assume this to be EPOCH in nanoseconds? But if so, what exactly does it describe? The time of arrival of a frame to the inference engine?

Could I please have an answer to these questions?

Please check the comments in nvdsmeta.h

  /** Holds a confidence value for the object, set by the inference
   component. confidence will be set to -0.1, if "Group Rectangles" mode of
   clustering is chosen since the algorithm does not preserve confidence
   values. Also, for objects found by tracker and not inference component,
   confidence will be set to -0.1 */
  gfloat confidence;

I’m using the Python bindings. How can I make them deliver confidence? And what is this timestamp?

I have tried to check this sample deepstream_python_apps/apps/deepstream-imagedata-multistream at master · NVIDIA-AI-IOT/deepstream_python_apps · GitHub. I got confidence values by specifying cluster-mode != 0 in my config. I understand, that a grouped detection indication cannot have a common confidence.

With cluster-mode=1 I’m getting a lot of detections, all drawn overlayed as red rectangles around me. cluster-mode=2 returns again to the well known single rectangle and there is a confidence != -0.1

What cluster mode would I have to chose in order to enable myself to walk over all detections and sort out everything below a given threshold, say 0.5 or 0.6?

Could one please explain the meaning of “ntp_timestamp”. I mean I know what a timestamp is, but what exactly describes this timestamp?


And again: Could please anybody explain, what that “ntp_timestamp” is?

This topic is about confidence, could you create a new topic for ntp_timestamp, we would like one topic track one question/issue.

Well, enough is enough.