Deepstream parallel pipeline video sink issue

[young2theMax]

Please provide complete information as applicable to your setup.

• Hardware Platform (Jetson / GPU) Jetson AGX Orin
• DeepStream Version 6.2
• JetPack Version (valid for Jetson only) 5.1
• TensorRT Version 8.5.2.2
• 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 am currently working on a parallel deepstream pipeline.
I refer to the following repo, https://github.com/NVIDIA-AI-IOT/deepstream_parallel_inference_app

My pipeline is like the picture below.

When I save the output video, let’s say output_1 and output_2, the issue rises.
The bounding box of the Yolov4 appears in the second video, output_2 with the bbox of Bodypose2D.

Is it not possible to encode two separate videos in the same pipeline?

[Fiona.Chen]

The configurations decide which model’s output to which stream. Can you post all your configurations?

[young2theMax]

Thanks for the reply Fiona.
Here are my config files.

[PGIE 0]

[property]
gpu-id=0
net-scale-factor=0.0039215697906911373
model-file=../model/Primary_Detector/resnet10.caffemodel
proto-file=../model/Primary_Detector/resnet10.prototxt
model-engine-file=../model/Primary_Detector/resnet10.caffemodel_b1_gpu0_int8.engine
labelfile-path=../model/Primary_Detector/labels.txt
int8-calib-file=../model/Primary_Detector/cal_trt.bin
force-implicit-batch-dim=1
batch-size=1
network-mode=1
num-detected-classes=4
interval=0
gie-unique-id=1
output-blob-names=conv2d_bbox;conv2d_cov/Sigmoid
#scaling-filter=0
#scaling-compute-hw=0

[class-attrs-all]
pre-cluster-threshold=0.2
eps=0.2
group-threshold=1

[PGIE 1]

[property]
gpu-id=0
net-scale-factor=1.0
model-color-format=0
model-engine-file=../model/unetres18_v4_pruned0.65_800_data.uff_b1_gpu0_fp16.engine
uff-file=../model/unetres18_v4_pruned0.65_800_data.uff
infer-dims=3;512;512
uff-input-order=0
uff-input-blob-name=data
## 0=FP32, 1=INT8, 2=FP16 mode
network-mode=2
num-detected-classes=4
interval=0
gie-unique-id=2
  
network-type=2
  
  
output-blob-names=final_conv/BiasAdd
segmentation-threshold=0.0
batch-size=1
  
[class-attrs-all]
roi-top-offset=0
roi-bottom-offset=0
detected-min-w=0
detected-min-h=0
detected-max-w=0
detected-max-h=0

[Fiona.Chen]

The model configurations are not for the pipeline control. Please post all app configurations, E.G. the configurations like deepstream_parallel_inference_app/tritonclient/sample/configs/apps/bodypose_yolo at master · NVIDIA-AI-IOT/deepstream_parallel_inference_app (github.com)

[young2theMax]

Well I made the pipeline with the code like the following link.
https://gist.github.com/crearo/a49a8805857f1237c401be14ba6d3b03

So the config files I use are those two.

[young2theMax]

Does nvstreamdemux have to attach in the pipeline?

[Fiona.Chen]

Yes. The nvstreamdemux should be in the pipeline and you need to implement the function of removing the sink branch when the corresponding source is removed.

[young2theMax]

Okay, I’ll try and share the result :) Thanks a lot!

[young2theMax]

I got one more related issue, well, it could be a simple question about queue.
Before adding streamdemux-streammux after tee(between tee-pgie), i added queue here just like the pic. below.

and the overlay issue(output of pgie_0 also shows on the output of pgie_1) does not happen on every frame.
So what are the reason for this? Just adding queue made the branch more independent.

[Fiona.Chen]

queue will not change any inferencing logic. queue (gstreamer.freedesktop.org)

Please make sure you have not change any code of GitHub - NVIDIA-AI-IOT/deepstream_parallel_inference_app: A project demonstrating how to use nvmetamux to run multiple models in parallel.. Please share your configurations of the app.

[young2theMax]

Okay I’ll try and share the result
Thanks :)

[yingliu]

Hi @young2theMax ,
Do you still need support for this topic? Or should we close it? Thanks.