Skip to content

Latest commit

 

History

History
35 lines (28 loc) · 1.04 KB

README.md

File metadata and controls

35 lines (28 loc) · 1.04 KB

Custom Gstreamer Element (GPU Usage)

Deepstream SDK custom gst-element. (NVMM memory --> GpuMat) Sample Repo to get buffer from nvmm memory and map it to gpumat without going to/from cpu. It uses cuda programming to draw a filled rectangle. (basic preprocessing)

System Environment

  • Hardware : Jetson Orin AGX
  • OS : Ubuntu 20.04
  • JetPack - 5.1.1
    • TensorRT - 8.5.2.2
    • Deepstream - 6.2
    • OpenCV - 4.6.0 + CUDA

Installation

git clone https://github.com/RajUpadhyay/Deepstream-Custom-Element
cd Deepstream-Custom-Element

Set your cuda version in the Makefile like this CUDA_VER?=11.4

Now run make and sudo make install

make
sudo make install

Please set up nvinfer by yourself or refer this repo

gst-launch-1.0 uridecodebin3 uri=file:///video.mp4 ! nvvidconv ! m.sink_0 nvstreammux name=m width=1920 height=960 batch-size=1 ! dscustomplugin x1=100 y1=100 x2=400 y2=300 ! nvinfer config-file-path=config_infer.txt ! nvdsosd ! nvvidconv ! fpsdisplaysink sync=0