How to keep old label when parsing value from output-tensor-meta=1 in sgie classifier (reinfer_interval>0)

Please provide complete information as applicable to your setup.

• Hardware Platform (Jetson / GPU) GPU
• DeepStream Version 6.1.1
• 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 would like to receive all sgie results at once and then post-process them and then add obj_label to obj_meta.
And rather than processing sgie every time, I want to proceed efficiently by setting the condition of (reinfer_interval>0).
However, when I run it, obj_label blinks, but when I looked at the open source, it seems to track past information with std::vector<NvDsInferAttribute> &attrList (like history variable).
Is there any way to maintain it in my situation?

if (pUserMeta->base_meta.meta_type == NVDSINFER_TENSOR_OUTPUT_META)
    NvDsClassifierMeta *classifier_meta =
        nvds_acquire_classifier_meta_from_pool (pBatchMeta);
    NvDsLabelInfo *label_info =
        nvds_acquire_label_info_meta_from_pool (pBatchMeta);
    classifier_meta->unique_component_id = unique_id;
    label_info->result_prob = xxx;
    pObjectMeta->text_params.display_text = g_strconcat (temp, " ", label_info->result_label, nullptr);
    g_free (temp);

    nvds_add_label_info_meta_to_classifier(classifier_meta, label_info);
    nvds_add_classifier_meta_to_object (pObjectMeta, classifier_meta);

  1. which sample are you referring to? what is the whole media pipeline?
  2. do you mean all object_label text sometimes disappears? did track id appear every time? can you use DeepStream sample deeptream-test2 to reproduce this issue?
  1. camera->pgie->tracker->sgie->osd->sink

There is no update from you for a period, assuming this is not an issue any more. Hence we are closing this topic. If need further support, please open a new one. Thanks.

  1. how did you set network-type? if network-type is 100, you need to parse inference results and add meta by yourself in probe function. please refer to sample deepstream-infer-tensor-meta-test.
  2. secondary-reinfer-interval means “Re-inference interval for objects, in frames”. please refer to doc.
  3. you can save trackid and sgie results in a map. then find classification result by tracker id. please refer to gst_nvinfer_process_objects.

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