Method

SPG: Unsupervised Domain Adaptation for 3D Object Detection via Semantic Point Generation [la] [SPG] [SPG]
https://arxiv.org/pdf/2108.06709v1.pdf

Submitted on 6 Mar. 2021 08:33 by
Qiangeng Xu (University of Southern California)

Running time:0.09 s
Environment:1 core @ 2.5 Ghz (Python + C/C++)

Method Description:
Semantic Point Generation
Parameters:
\alpha=0.2
Latex Bibtex:
@inproceedings{xu2021spg,
title={SPG: Unsupervised Domain Adaptation for
3D Object Detection via Semantic Point
Generation},
author={Xu, Qiangeng and Zhou, Yin and Wang,
Weiyue and Qi, Charles R. and Anguelov, Dragomir},
booktitle={Proceedings of the IEEE conference on
computer vision and pattern recognition (ICCV)},
pages={},
year={2021}
}

Detailed Results

Object detection and orientation estimation results. Results for object detection are given in terms of average precision (AP) and results for joint object detection and orientation estimation are provided in terms of average orientation similarity (AOS).


Benchmark Easy Moderate Hard
Car (Detection) 97.80 % 94.71 % 92.19 %
Car (Orientation) 40.02 % 38.73 % 38.52 %
Car (3D Detection) 90.50 % 82.13 % 78.90 %
Car (Bird's Eye View) 94.33 % 88.70 % 85.98 %
This table as LaTeX

2D object detection results.
This figure as: png eps pdf txt gnuplot



Orientation estimation results.
This figure as: png eps pdf txt gnuplot



3D object detection results.
This figure as: png eps pdf txt gnuplot



Bird's eye view results.
This figure as: png eps pdf txt gnuplot




eXTReMe Tracker