Method

Sem-Aug [la] [Sem-Aug]


Submitted on 22 Feb. 2022 10:17 by
Lin Zhao (Beijing Institute of Technology)

Running time:0.1 s
Environment:GPU @ 2.5 Ghz (Python)

Method Description:
Improving Camera-LiDAR Feature Fusion With Semantic
Augmentation for 3D Vehicle Detection
Parameters:
TBD
Latex Bibtex:
@ARTICLE{9830844, author={Zhao, Lin and Wang,
Meiling and Yue, Yufeng}, journal={IEEE Robotics
and Automation Letters}, title={Sem-Aug: Improving
Camera-LiDAR Feature Fusion With Semantic
Augmentation for 3D Vehicle Detection}, year=
{2022}, volume={7}, number={4}, pages={9358-
9365}, doi={10.1109/LRA.2022.3191208}}

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) 96.79 % 93.77 % 88.78 %
Car (Orientation) 96.78 % 93.69 % 88.69 %
Car (3D Detection) 89.41 % 80.77 % 75.90 %
Car (Bird's Eye View) 93.35 % 87.37 % 82.43 %
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