[deepstream_pose_estimation] Why is the network-type set to 100 in the config?

Accelerated Computing Intelligent Video Analytics DeepStream SDK
#1

• Hardware Platform (Jetson / GPU): Jetson TX2
• DeepStream Version: 5.0.1
• JetPack Version (valid for Jetson only): 4.4
• TensorRT Version: 7.1.3.0
• CUDA Version: 10.2

Why is the network-type set to 100 in the config?

According to the DeepStream 5.0.1 Plugin Manual, there is no value for network-type that says 100.
https://docs.nvidia.com/metropolis/deepstream/5.0.1/dev-guide/text/DS_plugin_gst-nvinfer.html

image
image1582×148 22.9 KB

Besides, according to this, I think network-type needs to be 0 or 2 to use TensorMeta.

In addition, output-tensor-meta is not set in config. Isn’t that 1?

#3

Hi,

network-type=100 indicates others.
For more details, please check the API document below:

https://docs.nvidia.com/metropolis/deepstream/sdk-api/Api/Infer/Infer.html?highlight=nvdsinfernetworktype_segmentation#_CPPv420NvDsInferNetworkType

Thanks.

#5