Method

Leveraging Semantics for End-to-End Learning of Disparity Estimation from Stereo Imagery [DispSegNet]


Submitted on 4 Sep. 2018 19:40 by
Junming Zhang (university of michigan)

Running time:0.9 s
Environment:GPU @ 2.5 Ghz (Python)

Method Description:
Leveraging Semantics for End-to-End Learning of
Disparity Estimation from Stereo Imagery. No
ground truth disparity map is needed.
Parameters:
Latex Bibtex:
@ARTICLE{dissegnet,
author={J. {Zhang} and K. A. {Skinner} and R.
{Vasudevan} and M. {Johnson-Roberson}},
journal={IEEE Robotics and Automation Letters},
title={DispSegNet: Leveraging Semantics for End-
to-End Learning of Disparity Estimation From
Stereo Imagery},
year={2019},
volume={4},
number={2},
pages={1162-1169},
keywords={Estimation;Semantics;Task
analysis;Image segmentation;Feature
extraction;Three-dimensional
displays;Convolution;Computer vision for
transportation;deep learning in robotics and
automation},
doi={10.1109/LRA.2019.2894913},
ISSN={2377-3766},
month={April},}

Detailed Results

This page provides detailed results for the method(s) selected. For the first 20 test images, the percentage of erroneous pixels is depicted in the table. We use the error metric described in Object Scene Flow for Autonomous Vehicles (CVPR 2015), which considers a pixel to be correctly estimated if the disparity or flow end-point error is <3px or <5% (for scene flow this criterion needs to be fulfilled for both disparity maps and the flow map). Underneath, the left input image, the estimated results and the error maps are shown (for disp_0/disp_1/flow/scene_flow, respectively). The error map uses the log-color scale described in Object Scene Flow for Autonomous Vehicles (CVPR 2015), depicting correct estimates (<3px or <5% error) in blue and wrong estimates in red color tones. Dark regions in the error images denote the occluded pixels which fall outside the image boundaries. The false color maps of the results are scaled to the largest ground truth disparity values / flow magnitudes.

Test Set Average

Error D1-bg D1-fg D1-all
All / All 4.20 16.97 6.33
All / Est 4.20 16.97 6.33
Noc / All 3.86 15.89 5.85
Noc / Est 3.86 15.89 5.85
This table as LaTeX

Test Image 0

Error D1-bg D1-fg D1-all
All / All 4.19 2.19 3.91
All / Est 4.19 2.19 3.91
Noc / All 4.14 2.19 3.87
Noc / Est 4.14 2.19 3.87
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 1

Error D1-bg D1-fg D1-all
All / All 3.41 10.30 4.18
All / Est 3.41 10.30 4.18
Noc / All 3.28 10.30 4.08
Noc / Est 3.28 10.30 4.08
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 2

Error D1-bg D1-fg D1-all
All / All 4.97 6.48 5.04
All / Est 4.97 6.48 5.04
Noc / All 4.45 6.48 4.55
Noc / Est 4.45 6.48 4.55
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 3

Error D1-bg D1-fg D1-all
All / All 5.60 19.83 6.92
All / Est 5.60 19.83 6.92
Noc / All 5.29 19.83 6.65
Noc / Est 5.29 19.83 6.65
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 4

Error D1-bg D1-fg D1-all
All / All 6.41 15.43 7.90
All / Est 6.41 15.43 7.90
Noc / All 5.69 15.43 7.34
Noc / Est 5.69 15.43 7.34
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 5

Error D1-bg D1-fg D1-all
All / All 13.38 17.05 13.71
All / Est 13.38 17.05 13.71
Noc / All 12.01 17.05 12.47
Noc / Est 12.01 17.05 12.47
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 6

Error D1-bg D1-fg D1-all
All / All 10.84 8.32 10.58
All / Est 10.84 8.32 10.58
Noc / All 10.96 8.32 10.67
Noc / Est 10.96 8.32 10.67
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 7

Error D1-bg D1-fg D1-all
All / All 1.38 12.16 3.49
All / Est 1.38 12.16 3.49
Noc / All 1.40 12.16 3.53
Noc / Est 1.40 12.16 3.53
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 8

Error D1-bg D1-fg D1-all
All / All 1.39 4.39 1.94
All / Est 1.39 4.39 1.94
Noc / All 1.39 4.39 1.94
Noc / Est 1.39 4.39 1.94
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 9

Error D1-bg D1-fg D1-all
All / All 1.52 5.44 2.52
All / Est 1.52 5.44 2.52
Noc / All 1.53 5.71 2.57
Noc / Est 1.53 5.71 2.57
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 10

Error D1-bg D1-fg D1-all
All / All 2.94 6.11 3.66
All / Est 2.94 6.11 3.66
Noc / All 2.98 6.11 3.70
Noc / Est 2.98 6.11 3.70
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 11

Error D1-bg D1-fg D1-all
All / All 2.20 2.27 2.22
All / Est 2.20 2.27 2.22
Noc / All 2.20 2.27 2.21
Noc / Est 2.20 2.27 2.21
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 12

Error D1-bg D1-fg D1-all
All / All 1.12 2.71 1.22
All / Est 1.12 2.71 1.22
Noc / All 1.13 2.71 1.24
Noc / Est 1.13 2.71 1.24
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 13

Error D1-bg D1-fg D1-all
All / All 1.63 3.22 1.83
All / Est 1.63 3.22 1.83
Noc / All 1.48 3.22 1.70
Noc / Est 1.48 3.22 1.70
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 14

Error D1-bg D1-fg D1-all
All / All 1.50 3.20 1.53
All / Est 1.50 3.20 1.53
Noc / All 1.37 3.20 1.40
Noc / Est 1.37 3.20 1.40
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 15

Error D1-bg D1-fg D1-all
All / All 4.19 5.64 4.32
All / Est 4.19 5.64 4.32
Noc / All 4.26 5.64 4.39
Noc / Est 4.26 5.64 4.39
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 16

Error D1-bg D1-fg D1-all
All / All 6.06 8.27 6.38
All / Est 6.06 8.27 6.38
Noc / All 5.65 8.27 6.04
Noc / Est 5.65 8.27 6.04
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 17

Error D1-bg D1-fg D1-all
All / All 2.00 1.52 1.95
All / Est 2.00 1.52 1.95
Noc / All 1.84 1.52 1.81
Noc / Est 1.84 1.52 1.81
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 18

Error D1-bg D1-fg D1-all
All / All 7.21 18.06 12.36
All / Est 7.21 18.06 12.36
Noc / All 7.06 18.06 12.34
Noc / Est 7.06 18.06 12.34
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 19

Error D1-bg D1-fg D1-all
All / All 1.66 4.41 1.97
All / Est 1.66 4.41 1.97
Noc / All 1.63 4.41 1.95
Noc / Est 1.63 4.41 1.95
This table as LaTeX

Input Image

D1 Result

D1 Error




eXTReMe Tracker