The KITTI Vision Benchmark Suite

Method

Hybrid Conditional Random Field [la] [HybridCRF]

Submitted on 21 Feb. 2017 08:04 by
Liang Xiao (National University of Defense Technology)

Running time:		1.5 s
Environment:		1 core @ 2.5 Ghz (C/C++)

Method Description:

Fuse Camera and LIDAR with Hybrid CRF model.

Parameters:

Latex Bibtex:

@article{XIAO2018HybridCRF,
title = "Hybrid conditional random field based
camera-LIDAR fusion for road detection",
journal = "Information Sciences",
volume = "432",
pages = "543 - 558",
year = "2018",
issn = "0020-0255",
doi =
"https://doi.org/10.1016/j.ins.2017.04.048",
url =
"http://www.sciencedirect.com/science/article/pii
/S0020025517307119",
author = "Liang Xiao and Ruili Wang and Bin Dai
and Yuqiang Fang and Daxue Liu and Tao Wu",
keywords = "Sensor fusion, Road detection,
Condition random field, Boosted decision tree"
}

Evaluation in Bird's Eye View

Benchmark	MaxF	AP	PRE	REC	FPR	FNR
UM_ROAD	90.99 %	85.26 %	90.65 %	91.33 %	4.29 %	8.67 %
UMM_ROAD	91.95 %	86.44 %	94.01 %	89.98 %	6.30 %	10.02 %
UU_ROAD	88.53 %	80.79 %	86.41 %	90.76 %	4.65 %	9.24 %
URBAN_ROAD	90.81 %	86.01 %	91.05 %	90.57 %	4.90 %	9.43 %

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.

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