Method

Vote3Deep: Fast Object Detection in 3D Point Clouds Using Efficient CNNs [la] [Vote3Deep]


Submitted on 15 Sep. 2016 16:50 by
Dushyant Rao (University of Oxford)

Running time:1.5 s
Environment:4 cores @ 2.5 Ghz (C/C++)

Method Description:
This is a computationally efficient approach to detecting
objects natively in 3D point clouds using convolutional neural
networks (CNNs). This is achieved by leveraging a feature-
centric voting scheme to implement novel convolutional layers
which explicitly exploit the sparsity encountered in the input.

Described in more detailed at: http://arxiv.org/abs/1609.06666
Parameters:
3 layer models for cyclists and pedestrians, 2 layer model for
cars.
Latex Bibtex:
@ARTICLE{Engelcke2016ARXIV,
author = {{Engelcke}, M. and {Rao}, D. and {Zeng Wang}, D.
and {Hay Tong}, C. and {Posner}, I.},
title = "{Vote3Deep: Fast Object Detection in 3D Point
Clouds Using Efficient Convolutional Neural Networks}",
journal = {ArXiv e-prints},
year = 2016}
}

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) 78.95 % 70.30 % 63.12 %
Pedestrian (Detection) 67.99 % 54.80 % 51.17 %
Cyclist (Detection) 78.41 % 68.82 % 62.50 %
This table as LaTeX


2D object detection results.
This figure as: png eps pdf txt gnuplot



2D object detection results.
This figure as: png eps pdf txt gnuplot



2D object detection results.
This figure as: png eps pdf txt gnuplot




eXTReMe Tracker