FAST-Dynamic-Vision is a system for detection and tracking dynamic objects with event and depth sensing.

FAST-Dynamic-Vision is a detection and trajectory estimation algorithm based on event and depth camera.

Related Paper: FAST-Dynamic-Vision: Detection and Tracking Dynamic Objects with Event and Depth Sensing, Botao He, Haojia Li, Siyuan Wu, Dong Wang, Zhiwei Zhang, Qianli Dong, Chao Xu, Fei Gao

Video Links: YouTube, Bilibili

• event-detector: Key modules with event processing, depth estimation, motion compensation and object detection.
• bullet_traj: modules for trajectory estimation in 3D world frame.

Requirement: Ubuntu 18.04 with ros-desktop-full installation; Ceres Solver; OpenCV 3; opencv_contrib

NOTION: If you are using Ubuntu 20.04 and failed to build this project with some linking error, please edit CMakeLists.txt and specify OpenCV 4 .

Step 1: Installation

sudo apt install libeigen3-dev build-essential libopencv-contrib-dev

Step 2: Clone the thie repo

git clone https://github.com/ZJU-FAST-Lab/FAST-Dynamic-Vision.git --branch=main

Step 3: build this project

cd FAST-Dynamic-Vision
catkin_make

Please clone branch dataset to use our demo dataset. It was recorded via DVXplorer, which provides 640x480 resolution.

This demo shows the performance of object detection and tracking algorithms.

source devel/setup.bash
roslaunch detector detection.launch

This demo shows detect and estimate the 3D trajectory of a throwing ball utilizing event and depth camera. The ground truth of the ball trajectory is captured by Vicon Motion Capture system.

source devel/setup.bash
roslaunch bullet_traj_est demo_traj_est_rviz.launch

The source code is released under GPLv3 license.

For any technical issues, please contact Haojia Li([email protected]), Botao He([email protected]), Siyuan Wu ([email protected]), or Fei GAO ([email protected]).

For commercial inquiries, please contact Fei GAO ([email protected]).