We are actively updating this repository (especially removing hard code and adding comments) to make it easy to use. If you have any questions, please open an issue. Thanks!
This is a ROS-based efficient online learning framework for object classification in 3D LiDAR scans, taking advantage of robust multi-target tracking to avoid the need for data annotation by a human expert. The system is only tested in Ubuntu 18.04 and ROS Melodic (compilation fails on Ubuntu 20.04 and ROS Noetic).
Please watch the videos below for more details.
[2023-01-11] Our evaluation results in KITTI 3D OBJECT DETECTION are ranked 276, 91, 125 in car, pedestrian and cyclist respectively.
[2023-01-11] Our evaluation results in KITTI 2D OBJECT DETECTION achieved rankings of 22 on pedestrian and 66 on cyclist!
Please read the readme file of each sub-package first and install the corresponding dependencies.
-
(Optional) Download the raw data from KITTI.
-
(Optional) Download the sample data for testing.
-
(Optional) Prepare a customized dataset according to the format of the sample data.
# launch/efficient_online_learning
# autoware_tracker/config/params.yaml
cd catkin_ws
source devel/setup.bash
roslaunch src/efficient_online_learning/launch/efficient_online_learning.launch
If you are considering using this code, please reference the following:
@inproceedings{efficient_online_learning,
author = {Rui Yang and Zhi Yan and Tao Yang and Yassine Ruichek},
title = {Efficient Online Transfer Learning for 3D Object Classification in Autonomous Driving},
booktitle = {Proceedings of the 2021 IEEE International Intelligent Transportation Systems Conference (ITSC)},
pages = {2950--2957},
year = {2021}
}