Point cloud clustering using YOLOv5 ROS

This is a ROS node for using YOLOv5 for real time point cloud segmentation. It uses segmentation provided in the official YOLOv5 repository. This project is adapted from Base ROS YOLOv5 implementation

Installation

Dependencies

This package is built and tested on Ubuntu 20.04 LTS and ROS Noetic with Python 3.8.

• Clone the packages to ROS workspace and install requirement for YOLOv5 submodule:

cd <ros_workspace>/src
git clone https://github.com/mats-robotics/detection_msgs.git
git clone --recurse-submodules https://github.com/tharaka27/ypc_ros.git 
cd ypc_ros/src/yolov5
pip3 install -r requirements.txt # install the requirements for yolov5

• Build the ROS package:

cd <ros_workspace>
catkin build ypc_ros # build the ROS package

• Make the Python script executable

cd <ros_workspace>/src/ypc_ros/src
chmod +x detect.py
chmod +x cluster.py
chmod +x camera_info.py

Basic usage

• Launch the node:

roslaunch ypc_ros ypc.launch

Using custom weights and dataset (Working)

• Put your weights into ypc_ros/src/yolov5
• Put the yaml file for your dataset classes into ypc_ros/src/yolov5/data
• Change related ROS parameters in yolov5.launch: weights, data

Reference

• YOLOv5 official repository: https://github.com/ultralytics/yolov5
• YOLOv3 ROS PyTorch: https://github.com/eriklindernoren/PyTorch-YOLOv3
• Darknet ROS: https://github.com/leggedrobotics/darknet_ros
• YOLOv5 ROS : https://github.com/mats-robotics/yolov5_ros