Method

MAFF-Net: Filter False Positive for 3D Vehicle Detection with Multi-modal Adaptive Feature Fusion [MAFF-Net(DAF-Pillar)]


Submitted on 24 Sep. 2020 04:08 by
Zehan Zhang (Shanghai Jiao Tong University)

Running time:0.04 s
Environment:1 core @ 2.5 Ghz (Python + C/C++)

Method Description:
See the paper.
Parameters:
See the paper.
Latex Bibtex:
@article{zhang2020maffnet,
title={MAFF-Net: Filter False Positive for 3D
Vehicle Detection with Multi-modal Adaptive Feature
Fusion},
author={Zhang, Zehan and Liang, Zhidong and Zhang,
Ming and Zhao, Xian and Yang Ming and Tan Wenming
and Pu, Shiliang},
journal={arXiv preprint arXiv:2009.10945},
year={2020}
}

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) 94.38 % 91.46 % 83.89 %
Car (Orientation) 94.17 % 90.78 % 83.17 %
Car (3D Detection) 85.52 % 75.04 % 67.61 %
Car (Bird's Eye View) 90.79 % 87.34 % 77.66 %
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