Method

Spatial Decoder Network [la] [SDN]


Submitted on 11 Oct. 2017 04:55 by
Jaeil Park (Seoul Robotics)

Running time:0.07 s
Environment:GPU @ 1.5 Ghz (Python)

Method Description:
Fast spatial prediction with 3D point
cloud
Parameters:
TBD
Latex Bibtex:
TBD

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) 73.18 % 48.40 % 46.18 %
Car (Orientation) 47.30 % 31.05 % 29.93 %
Car (3D Detection) 34.05 % 21.36 % 18.59 %
Car (Bird's Eye View) 73.82 % 55.29 % 48.48 %
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