Method

A Self-Supervised Permutation Approach to the Stereo Matching Problem [Permutation Stereo]


Submitted on 5 Mar. 2022 16:17 by
Pierre-Andre Brousseau (Universite de Montreal)

Running time:30 s
Environment:GPU @ 2.5 Ghz (Matlab)

Method Description:
This paper proposes a novel permutation
formulation to the stereo matching problem. Our
proposed approach introduces a permutation volume
which provides a natural representation of stereo
constraints and disentangles stereo matching from
monocular disparity estimation. It also has the
benefit of simultaneously computing disparity and
a confidence measure which provides explainability
and a simple confidence heuristic for occlusions.
In the context of self-supervised learning, the
stereo performance is validated for standard
testing datasets and the confidence maps are
validated through stereo-visibility. Results show
that the permutation volume increases stereo
performance and features good generalization
behaviour. We believe that measuring confidence is
a key part of explainability which is instrumental
to adoption of deep methods in critical stereo
applications such as autonomous navigation.
Parameters:
Our model is trained on the datasets at half
resolution on random image crops of size of 192 ×
32 pixels with a batch size of 2. No other data
augmentation is applied apart from random crops.
The conv blocks are as defined in [18] and their
Fig. 5 (right) with f=32. The convolution layers
apply fixed padding, have batch normalization and
have a Relu activation function. The
implementation is made with Mathematica[21] 12.3.
The λ is set to 10 and the symmetric normalization
has t=8 iterations. Networks are trained until
convergence with the Adam Optimizer[24] and a
learning rate of 1×10−3. The constant α is set to
0.85 as is customary [7] and τ is set to 0.1. The
models are trained on an RTX3090.
Latex Bibtex:
@inproceedings{brousseau2022permutation,
title={A Permutation Model for the Self-
Supervised Stereo Matching Problem},
author={Brousseau, Pierre-Andr{\'e} and Roy,
S{\'e}bastien},
booktitle={2022 19th Conference on Robots and
Vision (CRV)},
pages={122--131},
year={2022},
organization={IEEE}
}

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 5.53 15.47 7.18
All / Est 5.47 15.45 7.14
Noc / All 5.18 14.51 6.72
Noc / Est 5.13 14.49 6.68
This table as LaTeX

Test Image 0

Error D1-bg D1-fg D1-all
All / All 6.83 1.38 6.08
All / Est 6.82 1.38 6.07
Noc / All 6.82 1.38 6.06
Noc / Est 6.80 1.38 6.05
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 1

Error D1-bg D1-fg D1-all
All / All 4.33 10.73 5.05
All / Est 4.17 10.73 4.91
Noc / All 4.28 10.73 5.01
Noc / Est 4.12 10.73 4.87
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 2

Error D1-bg D1-fg D1-all
All / All 5.37 8.52 5.52
All / Est 5.36 8.52 5.52
Noc / All 4.98 8.52 5.15
Noc / Est 4.97 8.52 5.15
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 3

Error D1-bg D1-fg D1-all
All / All 6.41 12.91 7.01
All / Est 6.39 12.92 6.99
Noc / All 6.06 12.91 6.70
Noc / Est 6.04 12.92 6.69
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 4

Error D1-bg D1-fg D1-all
All / All 8.03 8.10 8.04
All / Est 8.00 8.10 8.02
Noc / All 7.46 8.10 7.57
Noc / Est 7.43 8.10 7.54
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 5

Error D1-bg D1-fg D1-all
All / All 11.79 9.39 11.57
All / Est 11.79 9.39 11.57
Noc / All 10.41 9.39 10.31
Noc / Est 10.41 9.39 10.31
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 6

Error D1-bg D1-fg D1-all
All / All 11.79 4.84 11.06
All / Est 11.79 4.84 11.06
Noc / All 11.53 4.84 10.81
Noc / Est 11.53 4.84 10.81
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 7

Error D1-bg D1-fg D1-all
All / All 2.72 9.31 4.01
All / Est 2.70 9.24 3.98
Noc / All 2.76 9.31 4.06
Noc / Est 2.74 9.24 4.03
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 8

Error D1-bg D1-fg D1-all
All / All 3.01 2.06 2.84
All / Est 3.01 2.06 2.83
Noc / All 3.01 2.06 2.83
Noc / Est 3.01 2.06 2.83
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 9

Error D1-bg D1-fg D1-all
All / All 3.72 2.94 3.52
All / Est 3.71 2.93 3.51
Noc / All 3.75 2.97 3.56
Noc / Est 3.74 2.96 3.55
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 10

Error D1-bg D1-fg D1-all
All / All 5.33 5.72 5.42
All / Est 5.27 5.72 5.38
Noc / All 5.40 5.72 5.47
Noc / Est 5.34 5.72 5.43
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 11

Error D1-bg D1-fg D1-all
All / All 2.69 2.31 2.62
All / Est 2.67 2.27 2.60
Noc / All 2.70 2.31 2.63
Noc / Est 2.69 2.27 2.61
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 12

Error D1-bg D1-fg D1-all
All / All 1.95 1.75 1.93
All / Est 1.94 1.75 1.93
Noc / All 1.97 1.75 1.95
Noc / Est 1.96 1.75 1.94
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 13

Error D1-bg D1-fg D1-all
All / All 2.29 3.33 2.42
All / Est 2.19 3.33 2.33
Noc / All 2.16 3.33 2.30
Noc / Est 2.13 3.33 2.28
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 14

Error D1-bg D1-fg D1-all
All / All 2.20 1.69 2.19
All / Est 2.20 1.58 2.19
Noc / All 2.04 1.69 2.03
Noc / Est 2.04 1.58 2.03
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 15

Error D1-bg D1-fg D1-all
All / All 4.57 4.20 4.54
All / Est 4.56 4.20 4.53
Noc / All 4.62 4.20 4.58
Noc / Est 4.61 4.20 4.57
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 16

Error D1-bg D1-fg D1-all
All / All 6.65 6.49 6.62
All / Est 6.56 6.49 6.55
Noc / All 6.39 6.49 6.41
Noc / Est 6.31 6.49 6.34
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 17

Error D1-bg D1-fg D1-all
All / All 2.79 2.56 2.77
All / Est 2.77 2.56 2.75
Noc / All 2.59 2.56 2.58
Noc / Est 2.56 2.56 2.56
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 18

Error D1-bg D1-fg D1-all
All / All 9.31 12.97 11.05
All / Est 9.26 12.97 11.03
Noc / All 9.22 12.97 11.02
Noc / Est 9.18 12.97 11.00
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 19

Error D1-bg D1-fg D1-all
All / All 1.97 5.15 2.33
All / Est 1.95 4.97 2.29
Noc / All 1.98 5.15 2.35
Noc / Est 1.96 4.97 2.31
This table as LaTeX

Input Image

D1 Result

D1 Error




eXTReMe Tracker