Description

This is a package combining darknet_ros and iai_kinect2 in order to get the 3D location of the objects detected.

It will automatically send tf transforms between the objects detected and kinect2_link.

System Requirements

• I only tested it on Ubuntu 20.04
• Remember to downgrade the gcc and g++ version of your system to 7 (my Ubuntu 20.04 is default 8)

Dependencies

• libfreenect2 (follow the official steps to install libfreenect2)
• python3-pcl (if you are using Ubuntu 18.04 with ROS Melodic, pip install python-pcl)

pip install python3-pcl

• ros_numpy (if you are using Ubuntu 18.04 with ROS Melodic, sudo apt-get install ros-melodic-ros-numpy)

sudo apt-get install ros-noetic-ros-numpy

• darknet_ros (This links to my forked repo. I modified it for supporting my RTX 30 laptop)
• iai_kinect2 (This links to my forked repo. I modified it for supporting Ubuntu 20.04)

Installation

cd <your_ws>/src

If you haven't cloned the dependencies packages yet, clone them first.

git clone --recursive https://github.com/0nhc/darknet_ros.git
git clone https://github.com/0nhc/iai_kinect2.git

Then, you can continue installing depth_yolo

git clone https://github.com/0nhc/depth_yolo.git
cd ..
catkin_make -DCMAKE_BUILD_TYPE=Release
source devel/setup.bash

Launch it

It will automatically launch darknet_ros node(yolo v3) and kinect2_bridge.

You can manually launch rviz to see the pointclouds and tf transforms.

roslaunch depth_yolo depth_yolo.launch