Method

Neighbor-Vote [Neighbor-Vote]


Submitted on 16 Apr. 2021 09:33 by
Xiaomeng Chu (University of Science and Technology of China)

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

Method Description:
Accepted by ACM Multimedia 2021
Parameters:
TBD
Latex Bibtex:
@misc{chu2021neighborvote,
title={Neighbor-Vote: Improving Monocular 3D
Object Detection through Neighbor Distance Voting},
author={Xiaomeng Chu and Jiajun Deng and Yao
Li and Zhenxun Yuan and Yanyong Zhang and Jianmin
Ji and Yu Zhang},
year={ACM MM 2021},
eprint={2107.02493},
archivePrefix={arXiv},
primaryClass={cs.CV}
}

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) 0.00 % 0.00 % 0.00 %
Car (3D Detection) 15.57 % 9.90 % 8.89 %
Car (Bird's Eye View) 27.39 % 18.65 % 16.54 %
This table as LaTeX

2D object detection 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