Method

CLCFNet[la] [CLCFNet]


Submitted on 3 Nov. 2020 08:55 by
Shuo Gu (Nanjing University of Science and Technology)

Running time:0.02 s
Environment:GPU @ 1.5 Ghz (Python)

Method Description:
LiDAR + Camera
about 23ms per frame
Parameters:
TBA
Latex Bibtex:
@inproceedings{GuYK21,
author = {Shuo Gu and
Jian Yang and
Hui Kong},
title = {A Cascaded LiDAR-Camera Fusion
Network for Road Detection},
booktitle = {{ICRA}},
publisher = {{IEEE}},
year = {2021}
}

Evaluation in Bird's Eye View


Benchmark MaxF AP PRE REC FPR FNR
UM_ROAD 95.65 % 89.49 % 95.31 % 96.00 % 2.15 % 4.00 %
UMM_ROAD 97.24 % 93.84 % 97.99 % 96.51 % 2.18 % 3.49 %
UU_ROAD 95.68 % 88.37 % 94.75 % 96.63 % 1.75 % 3.37 %
URBAN_ROAD 96.38 % 90.85 % 96.38 % 96.39 % 1.99 % 3.61 %
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

This figure as: png

This figure as: png

This figure as: png

This figure as: png

This figure as: png

This figure as: png

This figure as: png

This figure as: png

This figure as: png

This figure as: png

This figure as: png

This figure as: png

This figure as: png

This figure as: png

This figure as: png

This figure as: png

This figure as: png


eXTReMe Tracker