Method

SARPNET [SARPNET]


Submitted on 14 Nov. 2019 07:50 by
Yangyang Ye (Zhejiang University)

Running time:0.05 s
Environment:1 core @ 2.5 Ghz (Python + C/C++)

Method Description:
TBD
Parameters:
TBD
Latex Bibtex:
@article{ye2019sarpnet,
title={SARPNET: Shape Attention Regional Proposal
Network for LiDAR-based 3D Object Detection},
author={Ye, Yangyang and Chen, Houjin and Zhang,
Chi and Hao, Xiaoli and Zhang, Zhaoxiang},
journal={Neurocomputing},
year={2019},
publisher={Elsevier}
}

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) 96.07 % 93.21 % 88.09 %
Car (Orientation) 95.82 % 92.58 % 87.33 %
Car (3D Detection) 85.63 % 76.64 % 71.31 %
Car (Bird's Eye View) 92.21 % 86.92 % 81.68 %
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