Method

3D Object Proposals for Accurate Object Class Detection [st] [3DOP]
http://www.cs.toronto.edu/objprop3d/

Submitted on 11 Sep. 2015 03:37 by
Kaustav Kundu (University of Toronto)

Running time:3s
Environment:GPU @ 2.5 Ghz (Matlab + C/C++)

Method Description:
http://www.cs.toronto.edu/objprop3d/
Parameters:
Latex Bibtex:
@inproceedings{Chen2015NIPS,
title = {3D Object Proposals for Accurate Object Class
Detection},
author = {Xiaozhi Chen and Kaustav Kundu and Yukun Zhu
and Andrew Berneshawi and Huimin Ma and Sanja Fidler
and Raquel Urtasun},
booktitle = {NIPS},
year = {2015}
}

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) 92.96 % 89.55 % 79.38 %
Car (Orientation) 91.31 % 86.93 % 76.72 %
Pedestrian (Detection) 83.17 % 69.57 % 63.48 %
Pedestrian (Orientation) 74.22 % 61.48 % 55.89 %
Cyclist (Detection) 80.52 % 68.71 % 61.07 %
Cyclist (Orientation) 72.24 % 58.45 % 51.91 %
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



2D object detection results.
This figure as: png eps pdf txt gnuplot



Orientation estimation results.
This figure as: png eps pdf txt gnuplot



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