YOLOX-ROS Docker

en / ja

This repository provides an environment for running people_detection_ros2 in a Docker container.

The following packages this repository depends on :

• dbolya/yolact
• Megvii-BaseDetection/YOLOX
• Ar-Ray-code/YOLOX-ROS
• Rits-Interaction-Laboratory/shigure_tools

Required

---

• Docker

Building

---

Getting Dependent Packages :

bash startup.sh

Creating Docker Image :

docker build -t yolox_ros_docker .

Creating Docker container :

docker run -it --name ${container name} --gpus all --net host yolox_ros_docker:latest

After creating container

---

source /opt/ros/foxy/setup.bash
cd /ros2_ws
colcon build --symlink-install
. ./install/setup.bash

Install ros-foxy-v4l2-camera

apt-get update
apt install ros-foxy-v4l2-camera

Setting Weights

• File Location: Z:/aarimoto/best_ckpt1.zip
• Put the weight in /ros2/src/yolox_ros/weights

If you get an error message "TypeError: init() got an unexpected keyword argument 'encoding'"

---

• Go to directory

cd /opt/ros/foxy/share/ament_cmake_core/cmake/core

• Open File "package_xml_2_cmake.py"

vi package_xml_2_cmake.py

• Change "type=argparse.FileType('r', encoding='utf-8')" to "type=argparse.FileType('r')"

If you get an error message "ModuleNotFoundError: No module named '○○○○○' "

---

Install missing modules

pip install ○○○○○ 

Run node

---

Run the yolox-ros :

ros2 launch yolox_ros_py yolox_nano_torch_gpu_camera.launch.py