The KITTI Vision Benchmark Suite

Method

PLARD [la][at] [PLARD]
https://github.com/zhechen/PLARD

Submitted on 28 Nov. 2018 00:59 by
Zhe Chen (The University of Sydney)

Running time:		0.16 s
Environment:		GPU @ 2.5 Ghz (Python)

Method Description:

Image + LiDAR (waiting for publication)

Parameters:

40 epochs

Latex Bibtex:

@article{chen2019progressive,
title={Progressive LiDAR adaptation for road
detection},
author={Chen, Zhe and Zhang, Jing and Tao,
Dacheng},
journal={IEEE/CAA Journal of Automatica
Sinica},
volume={6},
number={3},
pages={693--702},
year={2019},
publisher={IEEE}
}

Evaluation in Bird's Eye View

Benchmark	MaxF	AP	PRE	REC	FPR	FNR
UM_ROAD	97.05 %	93.53 %	97.18 %	96.92 %	1.28 %	3.08 %
UMM_ROAD	97.77 %	95.64 %	97.75 %	97.79 %	2.48 %	2.21 %
UU_ROAD	95.95 %	95.25 %	96.25 %	95.65 %	1.21 %	4.35 %
URBAN_ROAD	97.03 %	94.03 %	97.19 %	96.88 %	1.54 %	3.12 %

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