Method

EBM3DOD baseline [EBM3DOD baseline]
https://github.com/fregu856/ebms_3dod

Submitted on 30 Oct. 2020 09:49 by
Fredrik Gustafsson (Uppsala University)

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

Method Description:
https://arxiv.org/abs/2012.04634
Parameters:
-
Latex Bibtex:
@article{gustafsson2020accurate,
title={Accurate 3D Object Detection using Energy-
Based Models},
author={Gustafsson, Fredrik K and Danelljan,
Martin and Schön, Thomas B},
journal={arXiv preprint arXiv:2012.04634},
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) 96.72 % 93.45 % 88.25 %
Car (Orientation) 96.31 % 92.70 % 87.44 %
Car (3D Detection) 88.80 % 79.52 % 72.30 %
Car (Bird's Eye View) 95.44 % 89.63 % 84.34 %
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