The KITTI Vision Benchmark Suite

Method

Fast multi-task CNN with orientation regression [at] [multi-task CNN]

Submitted on 18 Oct. 2018 17:32 by
Malte Oeljeklaus (TU Dortmund)

Running time:		25.1 ms
Environment:		GPU @ 2.0 Ghz (Python)

Method Description:

A convolutional neural network (CNN) comprised of a shared encoder stage and specific decoders for road segmentation and object detection. the detection stage is extended to predict the orientation of detected objects. The orientation estimate guides a consecutive 3D bounding box estimation based on analytic geometry.

Parameters:

Latex Bibtex:

@inproceedings{Oeljeklaus18,
title={A Fast Multi-Task CNN for Spatial Understanding of Traffic Scenes},
author={Oeljeklaus, Malte and Hoffmann, Frank and Bertram, Torsten},
booktitle={IEEE Intelligent Transportation Systems Conference},
year={2018}
}

Evaluation in Bird's Eye View

Benchmark	MaxF	AP	PRE	REC	FPR	FNR
UM_ROAD	85.95 %	81.28 %	77.40 %	96.64 %	12.86 %	3.36 %
UMM_ROAD	91.15 %	87.45 %	85.08 %	98.15 %	18.92 %	1.85 %
UU_ROAD	80.45 %	75.87 %	68.63 %	97.19 %	14.48 %	2.81 %
URBAN_ROAD	86.81 %	82.15 %	78.26 %	97.47 %	14.92 %	2.53 %

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