Ubuntu 18.04/20.04 CV环境配置(下)--手势识别TRTpose+Kinect DK人体骨骼识别

Posted Techblog of HaoWANG

tags:

篇首语:本文由小常识网(cha138.com)小编为大家整理,主要介绍了Ubuntu 18.04/20.04 CV环境配置(下)--手势识别TRTpose+Kinect DK人体骨骼识别相关的知识,希望对你有一定的参考价值。


https://github.com/Alex1114/TRT-Pose-ROShttps://github.com/Alex1114/TRT-Pose-ROS

GitHub - NVIDIA-AI-IOT/trt_pose: Real-time pose estimation accelerated with NVIDIA TensorRTReal-time pose estimation accelerated with NVIDIA TensorRT - GitHub - NVIDIA-AI-IOT/trt_pose: Real-time pose estimation accelerated with NVIDIA TensorRThttps://github.com/NVIDIA-AI-IOT/trt_pose

https://github.com/NVIDIA-AI-IOT/trt_pose_handhttps://github.com/NVIDIA-AI-IOT/trt_pose_hand


目录

1. 编译安装TRT Pose

 1.1 安装依赖

1.2  编译安装TRT Pose

1.3 编译& install ROS TRT hand Pose PKG

2 测试使用

3. 集成调试


1. 编译安装TRT Pose

 1.1 安装依赖

安装 torch2trt
git clone https://github.com/NVIDIA-AI-IOT/torch2trt
cd torch2trt
sudo python3 setup.py install --plugins

安装其他支持包
sudo pip3 install tqdm cython pycocotools
sudo apt-get install python3-matplotlib

1.2  编译安装TRT Pose

git clone https://github.com/NVIDIA-AI-IOT/torch2trt
cd torch2trt
python setup.py install
# 加上pulgin会失败
 
pip install tqdm cython pycocotools
sudo apt-get install python3-matplotlib
 
git clone https://github.com/NVIDIA-AI-IOT/trt_pose
cd trt_pose
python setup.py install
 
# 安装jupyter lab
pip install jupyterlab
 
cd tasks/humanpose
jupyter lab
# 然后可以运行demo了

1.3 编译& install ROS TRT hand Pose PKG

  • Build and Install ros_trt_pose package
 catkin build
 source install/local_setup.sh

2 测试使用

硬件:realsense 435i + kinect DK + PC

Hand gesture recoginition (hand pose classification) 

roslaunch trt_pose_hand trt_gesture_classification.launch

3. 集成调试

./auto_start.sh

#!/bin/bash
# ALL Right reserved HAO,WANG
# @autor:       haowanghk@gmail.com
# @time:        2022-07-04-10:52
# @position:    JIhualab,China



# aim to launch realsense and hand pose TRT NVIDAIA Engine
# Master node has auto stated in dual_arm robots computer

# 1. load the cv_bridge pkg for python3.6

# Master node is in dual_arm robots
# launch k4a and k4abt  and node transfer 

# 打开终端窗口
gnome-terminal --tab --title="TRT Hand Pose" -- bash -c "cd /home/haowang/catkin_ws/install/;
				source setup.bash;
				source setup.bash;
				roslaunch trt_pose_hand trt_gesture_classification.launch;exec bash" &

sleep 1.5
echo "-----TRT and Realsense Driver controller node starting Success!--"


gnome-terminal --tab --title="roslaunch k4abt" -- bash -c "roslaunch azure_kinect_ros body_skeleton_detection.launch;exec bash" &

sleep 1.5
echo "-----K4a devices and k4abt node with RVIZ starting Success!--"


gnome-terminal --tab --title="rosrun rqt control cmd " -- bash -c "rqt;exec bash" &
sleep 1
echo "--  RQT Node starting Success!--"

gnome-terminal --tab --title="ROS ZED Camera HTTP Web Server " -- bash -c "rosrun web_video_server web_video_server;exec bash" &
sleep 1
echo "--  ROS HTTP Web Server has started!--"

wait


exit 0
~   

测试结果(工程代码不开源)

 

以上是关于Ubuntu 18.04/20.04 CV环境配置(下)--手势识别TRTpose+Kinect DK人体骨骼识别的主要内容,如果未能解决你的问题,请参考以下文章

Ubuntu 18.04/20.04 CV环境配置(中):Tensorrt + Pytorch安装配置

Ubuntu 18.04/20.04 CV环境配置(中):Tensorrt + Pytorch安装配置

Ubuntu 18.04/20.04 CV环境配置(上):CUDA11.1 + cudnn安装配置

Ubuntu 18.04/20.04 CV环境配置(上):CUDA11.1 + cudnn安装配置

Ubuntu 18.04/20.04 CV环境配置(上):CUDA11.1 + cudnn安装配置

Ubuntu 18.04/20.04 CV环境配置(上):CUDA11.1 + cudnn安装配置