Method

Frustum ConvNet: Sliding Frustums to Aggregate Local Point-Wise Features for Amodal 3D Object De[la] [F-ConvNet]
https://github.com/zhixinwang/frustum-convnet

Submitted on 20 Feb. 2019 10:11 by
Zhixin Wang (South China University of Technology )

Running time:0.47 s
Environment:GPU @ 2.5 Ghz (Python + C/C++)

Method Description:
See the paper.
(Runtime includes the 2D and 3D detectors. As for
car category, 0.4s for 2D detection on one image
and 0.07s for 3d detection on 32 region proposals.
)

Parameters:
See the paper.
Latex Bibtex:
@inproceedings{wang2019frustum,
title={Frustum ConvNet: Sliding Frustums to
Aggregate Local Point-Wise Features for Amodal 3D
Object Detection},
author={Wang, Zhixin and Jia, Kui},
booktitle={IROS},
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) 95.85 % 92.19 % 80.09 %
Car (Orientation) 95.81 % 91.98 % 79.83 %
Car (3D Detection) 87.36 % 76.39 % 66.69 %
Car (Bird's Eye View) 91.51 % 85.84 % 76.11 %
Pedestrian (Detection) 83.63 % 72.91 % 67.18 %
Pedestrian (Orientation) 75.19 % 63.87 % 58.57 %
Pedestrian (3D Detection) 52.16 % 43.38 % 38.80 %
Pedestrian (Bird's Eye View) 57.04 % 48.96 % 44.33 %
Cyclist (Detection) 86.75 % 78.05 % 68.12 %
Cyclist (Orientation) 86.39 % 76.71 % 66.92 %
Cyclist (3D Detection) 81.98 % 65.07 % 56.54 %
Cyclist (Bird's Eye View) 84.16 % 68.88 % 60.05 %
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



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



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