Jetson Nano Creating a Human Pose Estimation Application with NVIDIA DeepStream

参考:DeepStream-l4t

1. Jetson Nano 安装 DeepStream-l4t

---

1. docker pull:拉取镜像

docker pull nvcr.io/nvidia/deepstream-l4t:5.0.1-20.09-samples

2. Run the container:运行镜像
   2.1 Allow external applications to connect to the host’s X display:（允许外部应用程序连接到主机的X显示器）

   xhost +
   

   2.2 Run the docker container using the nvidia-docker (use the desired container tag in the command line below):启动容器

   sudo docker run -it --net=host --runtime nvidia  -e DISPLAY=$DISPLAY -w /opt/nvidia/deepstream/deepstream-5.0 -v /tmp/.X11-unix/:/tmp/.X11-unix nvcr.io/nvidia/deepstream-l4t:5.0.1-20.09-samples
   

2. Creating a Human Pose Estimation Application with NVIDIA DeepStream

参考:Creating a Human Pose Estimation Application with NVIDIA DeepStream
环境依赖:

1. DeepStream SDK 5.0 : 上面已经安装
2. CUDA 10.2: jetsonnano预装
3. TensorRT 7.x: jetsonnano预装

---

Step 1: Write the post-processing code required for your model

下载 deepstream_pose_estimation代码 NVIDIA-AI-IOT/deepstream_pose_estimation

git clone https://github.com/NVIDIA-AI-IOT/deepstream_pose_estimation.git

Step 2: Download the human pose estimation model and convert it to ONNX

下载 NVIDIA-AI-IOT/trt_pose 代码 NVIDIA-AI-IOT/trt_pose

git clone https://github.com/NVIDIA-AI-IOT/trt_pose.git

Step 3: Replace the OSD library in the DeepStream install directory

Step 4: Edit the DeepStream configuration file

Step 5: Edit the makefile to include platform-specific build flags

Step 6: Compile and run the DeepStream app