The KITTI Vision Benchmark Suite

Method

TEDNet [la] [TEDNet]
https://github.com/martin-bayon/TEDNet

Submitted on 31 Jan. 2023 08:43 by
Martín Bayón (Universidad de León)

Running time:		0.09 s
Environment:		GPU @ 2.5 Ghz (Python)

Method Description:

Please, read our article paper for further
details

Parameters:

Please, read our article paper for further
details

Latex Bibtex:

@article{10.1093/jigpal/jzae048,
author = {Bayón-Gutiérrez, Martín and García-
Ordás, María Teresa and Alaiz Moretón, Héctor and
Aveleira-Mata, Jose and Rubio-Martín, Sergio and
Benítez-Andrades, José Alberto},
title = "{TEDNet: Twin Encoder Decoder Neural
Network for 2D Camera and LiDAR Road Detection}",
journal = {Logic Journal of the IGPL},
pages = {jzae048},
year = {2024},
month = {05},
issn = {1367-0751},
doi = {10.1093/jigpal/jzae048},
url =
{https://doi.org/10.1093/jigpal/jzae048},
eprint =
{https://academic.oup.com/jigpal/advance-article-
pdf/doi/10.1093/jigpal/jzae048/57502710/jzae048.pd
f},
}

Evaluation in Bird's Eye View

Benchmark	MaxF	AP	PRE	REC	FPR	FNR
UM_ROAD	94.24 %	92.43 %	93.45 %	95.04 %	3.04 %	4.96 %
UMM_ROAD	95.94 %	95.31 %	96.21 %	95.67 %	4.14 %	4.33 %
UU_ROAD	92.78 %	91.08 %	91.86 %	93.71 %	2.71 %	6.29 %
URBAN_ROAD	94.62 %	93.05 %	94.28 %	94.96 %	3.17 %	5.04 %

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