ConDA: Unsupervised Domain Adaptation for LiDAR Segmentation via Regularized Domain Concatenation
Lingdong Kong 1,*,
Niamul Quader 2,
Venice Erin Liong 2,
Hanwang Zhang 1
1Nanyang Technological University, 2Motional
*Work done as an autonomous vehicle intern at Motional
This repository contains the PyTorch implementation of ConDA: Unsupervised Domain Adaptation for LiDAR Segmentation via Regularized Domain Concatenation.
This is not an official Motional product.
ConDA can process raw point clouds for unsupervised domain adaptation (UDA) in LiDAR segmentation. It also supports other domain adaptation settings under annotation scarcity, such as semi-supervised domain adaptation (SSDA) and weakly-supervised domain adaptation (WSDA).
Coming soon.
Coming soon.
@article{kong2021conda,
title = {ConDA: Unsupervised domain adaptation for LiDAR segmentation via regularized domain concatenation},
author = {Lingdong Kong and Niamul Quader and Venice Erin Liong and Hanwang Zhang},
journal = {arXiv preprint arXiv:2111.15242},
year = {2021}
}
ConDA is provided under the terms of the MIT license. For more details, see LICENSE.