Method

BS3D [BS3D]


Submitted on 8 Feb. 2019 10:30 by
Nils Gählert (Mercedes-Benz, R&D)

Running time:22 ms
Environment:Titan Xp

Method Description:
TBD
Parameters:
TBD
Latex Bibtex:
@inproceedings{gahlert2019beyond,
title={Beyond Bounding Boxes: Using Bounding
Shapes for Real-Time 3D Vehicle Detection from
Monocular RGB Images},
author={Gählert, Nils and Wan, Jun-Jun and
Weber, Michael and Zöllner, J. Marius and
Franke, Uwe and Denzler, Joachim},
booktitle={2019 IEEE Intelligent Vehicles
Symposium (IV)},
year={2019},
organization={IEEE}
}

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) 95.35 % 82.72 % 70.01 %
Car (Orientation) 94.66 % 81.22 % 68.39 %
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




eXTReMe Tracker