Method

Cascaded Sliding Window Based Real-Time 3D Region Proposal for Pedestrian Detection [la] [CSW3D]


Submitted on 1 Oct. 2019 17:02 by
Jun Hu (NUDT)

Running time:0.03 s
Environment:4 cores @ 2.5 Ghz (C/C++)

Method Description:
In The Paper.
Parameters:
In The Paper.
Latex Bibtex:
@inproceedings{hu2019csw3d,
title={Cascaded Sliding Window Based Real-Time
3D Region Proposal for Pedestrian Detection},
author={Jun, Hu and Tao, Wu and Hao, Fu and Zhiyu, Wang and Kai, Ding},
booktitle={ROBIO},
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
Pedestrian (Detection) 53.76 % 41.50 % 37.25 %
Pedestrian (3D Detection) 33.75 % 26.64 % 23.34 %
Pedestrian (Bird's Eye View) 49.27 % 37.96 % 33.83 %
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