Closed Abulala10 closed 1 year ago
It's better to create your own code instead to edit the deepstream-app
. See: https://github.com/NVIDIA-AI-IOT/deepstream_python_apps
Hello @marcoslucianops , I also have a similar issue . But for my case , I am using deepstream_app in c . Any suggestions regarding this without python-bindings (I do not want to switch to python) .
You can use the C/C++ apps. They are located in /opt/nvidia/deepstream/deepstream/sources/apps/sample_apps
.
It's better to create your own code instead to edit the
deepstream-app
. See: https://github.com/NVIDIA-AI-IOT/deepstream_python_apps
Okay and do you have any idea as to what part of code I should edit, if I am using the C/C++ app ?
For the test apps (test1, test2, etc), see the pad_buffer_probe
functions. Take a look on https://github.com/marcoslucianops/DeepStream-Yolo#extract-metadata to see strcuture of the NvDsFrameMeta
and NvDsObjMeta
.
Great @marcoslucianops I was able to do some basic changes and run the app , but I am facing issues while trying to display my confidence scores while inferencing . Could please guide me how to display the confidence score on the screen ?
Without tracker
snprintf(obj_meta->text_params.display_text, MAX_LABEL_SIZE, "%s (%.2f)", obj_meta->obj_label, obj_meta->confidence);
With tracker
snprintf(obj_meta->text_params.display_text, MAX_LABEL_SIZE, "%s %ld (%.2f)", obj_meta->obj_label, obj_meta->object_id, obj_meta->confidence);
Without tracker
snprintf(obj_meta->text_params.display_text, MAX_LABEL_SIZE, "%s (%.2f)", obj_meta->obj_label, obj_meta->confidence);
With tracker
snprintf(obj_meta->text_params.display_text, MAX_LABEL_SIZE, "%s %ld (%.2f)", obj_meta->obj_label, obj_meta->object_id, obj_meta->confidence);
Thank you @marcoslucianops !!! It works !
Cheers @marcoslucianops , I am able to display the conf values on the screen .
Thank you !!
I understood that we have to edit the process_meta() function in deepstream_app.c to show the confidence value on screen, but what exactly do we have to edit ?