andreashovaldt / yolov8_ros2

ROS2 Package for instance segmentation using the YoloV8 segmentation model and a Intel Realsense D435 camera

yolov8_ros2

Used in the 5th semester Robotics LLM-Planner-for-Bimanual-object-mnipulation project. Package provides functionalities for using the Intel Realsense D435 camera combined with the yoloV8 object segmentation model implemented using ROS2.

Required libraries

Python3

pip install open3d
pip install ultralytics

Quick Start

Install colcon and rosdep, then build this repository:

mkdir -p ~/ros2_ws/src && cd ~/ros2_ws/src
git clone https://github.com/andreasHovaldt/yolov8_ros2.git
cd ~/ros2_ws
rosdep install --from-paths src --ignore-src -r -y
colcon build --symlink-install
source ~/ros2_ws/install/setup.bash

Run code

ros2 launch yolov8_ros2 camera_yolo.launch.py

This launches the camera node, and the yolov8 node. The image prediction results of the yolov8 segmentation are published to the topic: /yolo/prediction/image. These results are also published in a json formatted string to the topic: /yolo/prediction/item_dict.