Method

Monocular 3D Object Detection with Decoupled Structured Polygon Estimation and Height-Guided Depth [Decoupled-3D]
[Anonymous Submission]

Submitted on 18 Oct. 2019 13:13 by
[Anonymous Submission]

Running time:0.08 s
Environment:GPU @ 2.5 Ghz (C/C++)

Method Description:
An efficient monocular 3D object detection framework that
decomposes the complicated 3D detection problem into a
structured polygon prediction task and a following depth
recovery task.
Parameters:
height = 1.46m
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) 87.78 % 67.92 % 54.53 %
Car (Orientation) 87.34 % 67.23 % 53.84 %
Car (3D Detection) 11.08 % 7.02 % 5.63 %
Car (Bird's Eye View) 23.16 % 14.82 % 11.25 %
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