Method

[la] Real-Time 3D Vehicle Detection in LiDAR Point Cloud for Autonomous Driving [RT3D]
[Anonymous Submission]

Submitted on 11 May. 2018 15:02 by
[Anonymous Submission]

Running time:0.09 s
Environment:GPU @ 1.8Ghz

Method Description:
RoI generated from BEV
and pre-RoI-pooling conv are used to boost speed
Parameters:
lr = 0.001
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) 49.96 % 39.71 % 41.47 %
Car (Orientation) 24.23 % 18.98 % 20.56 %
Car (3D Detection) 23.49 % 21.27 % 19.81 %
Car (Bird's Eye View) 54.68 % 42.10 % 44.05 %
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




eXTReMe Tracker