ROS YOLO MODEL ROS2 Package

A ROS 2 package, written in C++, leveraging YOLOv5 for object detection in various input sources (camera, video, image). This project integrates deep learning-based detection into the ROS 2 framework, providing flexibility for robotics applications. Developed and tested in a WSL (Windows Subsystem for Linux) environment.

Features

• Live Camera Detection: Detection with camera streams.
• Video & Image Detection: Process pre-recorded videos or individual images.
• Customizable Classes: Detects 22 object classes, including PASCAL VOC 2012 dataset objects and custom classes like university student and employee ID cards.
• ROS Topics Integration: Publishes detection results on /detected_objects topic. The implementation is designed to run under ROS 2 Humble.

Clone the packages

It is assumed that the workspace is ~/ros2_ws/.

cd ~/ros2_ws/src

git clone https://github.com/SlayM4yd4y/ros_yolo_model

git clone https://github.com/ultralytics/yolov5  
cd yolov5
pip install -r requirements.txt  

Build this ROS 2 package

It is assumed that the workspace is still ~/ros2_ws/.

cd ~/ros2_ws

colcon build --packages-select ros_yolo_model --symlink-install

Run this ROS 2 package

For more details visit the /wiki of this package.

Diagram

graph LR;

video_publisher([ /video_publisher_node]):::red --> image[ /image<br/>sensor_msgs/Image]:::light
image --> detector([ /detector_node]):::red
detector --> detected_objects[ /detected_objects<br/>std_msgs/String]:::light

classDef light fill:#34aec5,stroke:#152742,stroke-width:2px,color:#152742  
classDef dark fill:#152742,stroke:#34aec5,stroke-width:2px,color:#34aec5
classDef white fill:#ffffff,stroke:#152742,stroke-width:2px,color:#152742
classDef red fill:#ef4638,stroke:#152742,stroke-width:2px,color:#fff

Loading