Method

Clipping Gradient and Local Module Network [ClGLMNet]


Submitted on 25 Aug. 2020 11:52 by
Seunghwan Byun (Soongsil University)

Running time:0.03 s
Environment:GPU @ 2.5 Ghz (Python)

Method Description:
Camera + LiDAR
Parameters:
137
Latex Bibtex:

Evaluation in Bird's Eye View


Benchmark MaxF AP PRE REC FPR FNR
UM_ROAD 86.86 % 89.93 % 87.97 % 85.78 % 5.34 % 14.22 %
UMM_ROAD 91.95 % 94.53 % 93.45 % 90.49 % 6.97 % 9.51 %
UU_ROAD 84.94 % 88.63 % 87.02 % 82.95 % 4.03 % 17.05 %
URBAN_ROAD 88.73 % 91.68 % 90.61 % 86.93 % 4.96 % 13.07 %
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

This figure as: png eps pdf

This figure as: png eps pdf

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

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