Method

RenderAndCompare [RaC]
[Anonymous Submission]

Submitted on 21 May. 2017 21:42 by
[Anonymous Submission]

Running time:1s s
Environment:GPU @ 1.0 Ghz (C/C++)

Method Description:
Amodal 3D Object Instance Reconstruction via
Render-and-Compare network
Parameters:
More Later
Latex Bibtex:

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) 90.02 % 89.39 % 80.29 %
Car (Orientation) 89.98 % 89.25 % 80.07 %
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