EVIMO Pipeline Setup on Host OS

A Docker container is provided which can be used for most tasks with the dataset. However, instructions for setting up a ROS workspace on the host OS are provided below.

  1. The code was tested on Ubuntu 18.04 and 20.04 and consists of multiple tools for visualization, calibration and ground truth generation.

  2. While the released dataset is self-contained, in order to generate ground truth from raw .bag recordings you will need ROS installed on your system:

Setup

  1. Make sure ROS is installed; This project also relies on OpenCV and (to a lesser degree) PCL.

  2. If you do not have a catkin workspace set up:

  3. Download the cognifli code (see project page for more details):

cd ~/
git clone https://github.com/ncos/cognifli
  1. Run the INSTALL.py configuration tool to set up your catkin workspace:

cd ~/cognifli/contrib
./INSTALL.py
  1. Download packages with DVS/Vicon ROS message descriptions: link

  2. OR Download DVS and Vicon packages from their repositories:

cd ~/cognifli/src
git clone https://github.com/catkin/catkin_simple.git
git clone https://github.com/uzh-rpg/rpg_dvs_ros.git
git clone https://github.com/KumarRobotics/vicon

Note: only message description parts of the packages are needed; you can remove the rest of the package in case you get compilation errors.

  1. Download Evimo source code:

cd ~/cognifli/src
git clone https://github.com/better-flow/evimo
  1. Build the project:

cd ~/cognifli
catkin_make

Note: We use DVS event message format for both Samsung and Prophesee recordings.