Method

SPANet: Spatial and part-aware aggregation network for 3D object detection [SPANet]


Submitted on 30 Aug. 2021 13:15 by
Yangyang Ye (Zhejiang University)

Running time:0.06 s
Environment:1 core @ 2.5 Ghz (C/C++)

Method Description:
It utilizes a spatial aggregation network to remedy
the inhomogeneity of LiDAR point clouds, and
embodies a part-aware aggregation network that
learns the statistic shape priors of objects.
Parameters:
TBD
Latex Bibtex:
@inproceedings{ye2021spanet,
title={SPANet: Spatial and Part-Aware Aggregation Network
for 3D Object Detection},
author={Ye, Yangyang},
booktitle={Pacific Rim International Conference on Artificial
Intelligence},
pages={308--320},
year={2021},
organization={Springer}

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) 96.54 % 95.46 % 90.47 %
Car (Orientation) 96.31 % 95.03 % 89.99 %
Car (3D Detection) 91.05 % 80.34 % 74.89 %
Car (Bird's Eye View) 95.59 % 91.59 % 86.53 %
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