YOLOv8 with ROS2

This repository contains the implementation of YOLOv8 integrated with ROS 2 (Robot Operating System 2) for real-time object detection. YOLOv8 is the version of the “You Only Look Once” (YOLO) family of models, designed for efficient and accurate object detection.

Features

• YOLOv8 integration: Utilize the latest advancements in object detection using the YOLOv8 architecture.
• ROS 2 support: Communicate and control through ROS 2, making it easier to use in robotics applications.
• Real-time detection: Optimized for real-time object detection on robotic platforms.
• Extensible design: Easily customizable for different detection tasks and sensors.

Requirements

To run this project, you need to have the following installed:

• Python 3.8+
• ROS 2 (Foxy, Galactic, or later)
• PyTorch
• OpenCV
• TensorRT (optional for GPU acceleration)

Installation

1. Clone this repository to your local machine:

   git clone https://github.com/enginksz/YOLOV8_ROS2.git
   cd yolov8_ros2
   

2. Install dependencies:

   pip install ultralytics
   sudo apt install gazebo
   sudo apt install python3-colcon-common-extensions
   sudo apt install ros-foxy-gazebo-ros
   

3. Ensure that ROS 2 is properly installed and sourced on your machine.

Running YOLOv8 with ROS 2

1. Colcon build:

   cd yolov8_ros2
   colcon build
   source yolov8_ros2/install/setup.bash
   

2. Launch ROS 2 nodes:

   ros2 launch yolov8_gazebo yolov8_ros2_launch.py
   

   Open a second terminal for recognition:

   ros2 launch yolov8_ros2_recognition launch_yolov8.launch.py
   

3. Start detection by publishing images to the corresponding ROS 2 topic.

Usage

ROS 2 Topics

• /rgb_cam/image_raw: Input image topic for object detection.
• /inference_result_cv2: Output topic img_pub
• /inference_result: Output topic img_pub
• /Yolov8_Inference: Output topic yolov8_pub

Contributing

Feel free to submit issues, create pull requests, or fork this repository to improve and extend the project.