The KITTI Vision Benchmark Suite

Method

Road Estimation with Sparse 3D Points From Velodyne [la] [RES3D-Velo]

Submitted on 28 May. 2014 03:34 by
Patrick Shinzato (Mobile Robotic Laboratory, ICMC - USP)

Running time:		0.36 s
Environment:		1 core @ 2.5 Ghz (C/C++)

Method Description:

This method is based on fusion of a sparse and unstructured 3D point clouds and images. The main idea is use spatial-relationship in image perspective view combined to real 3D metric values to determine if a point corresponds to an obstacle or not. After that, conbinations of polar histograms are used to generate a confidence map that represents the road area in the image.

Parameters:

\theta = 77.0

Latex Bibtex:

@inproceedings{Shinzato2014IV,
author = "Patrick Yuri Shinzato and Denis Fernando Wolf and Christoph Stiller",
title = "Road Terrain Detection: Avoiding Common Obstacle Detection Assumptions Using Sensor Fusion",
booktitle = "Intelligent Vehicles Symposium (IV)",
year = "2014",
}

Evaluation in Bird's Eye View

Benchmark	MaxF	AP	PRE	REC	FPR	FNR
UM_ROAD	83.81 %	73.95 %	78.56 %	89.80 %	11.16 %	10.20 %
UMM_ROAD	90.60 %	85.38 %	85.96 %	95.78 %	17.20 %	4.22 %
UU_ROAD	83.63 %	72.58 %	77.38 %	90.97 %	8.67 %	9.03 %
URBAN_ROAD	86.58 %	78.34 %	82.63 %	90.92 %	10.53 %	9.08 %

This table as LaTeX

Behavior Evaluation

Benchmark	PRE-20	F1-20	HR-20	PRE-30	F1-30	HR-30	PRE-40	F1-40	HR-40

This table as LaTeX

Road/Lane Detection

The following plots show precision/recall curves for the bird's eye view evaluation.

This figure as: png eps pdf

Distance-dependent Behavior Evaluation

The following plots show the F1 score/Precision/Hitrate with respect to the longitudinal distance which has been used for evaluation.

Visualization of Results

The following images illustrate the performance of the method qualitatively on a couple of test images. We first show results in the perspective image, followed by evaluation in bird's eye view. Here, red denotes false negatives, blue areas correspond to false positives and green represents true positives.

This figure as: png