Method

4D-IteraStereo: Stereo Matching via Cost Aggregation and Iterative Optimization [4D-IteraStereo]
[Anonymous Submission]

Submitted on 5 Jan. 2025 13:14 by
[Anonymous Submission]

Running time:0.3 s
Environment:GPU @ 1.5 Ghz (Python)

Method Description:
The 4D-IteraStereo method introduces a novel
approach to stereo matching by optimizing cost
aggregation and enhancing matching accuracy,
particularly in texture-less regions. Unlike
traditional methods that suffer from redundant
features and lack geometric information, 4D-
IteraStereo builds a lightweight 4D cost volume
and uses parallel asymmetric and dilated
convolutions for efficient cost aggregation. This
innovative design reduces computational load and
provides better regularization of the cost volume.
Furthermore, by integrating global and scene-prior
knowledge, the method effectively improves
accuracy in challenging texture-less areas and
fine structures. Experimental results confirm that
the proposed method significantly improves the
accuracy of stereo measurements in texture-less
scenes.
Parameters:
NAME: AdamW
LR: &lr 0.0002
WEIGHT_DECAY: 1.0e-05
EPS: 1.0e-08
Latex Bibtex:

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 1.60 2.48 1.75
All / Est 1.60 2.48 1.75
Noc / All 1.46 2.43 1.62
Noc / Est 1.46 2.43 1.62
This table as LaTeX

Test Image 0

Error D1-bg D1-fg D1-all
All / All 1.42 1.17 1.39
All / Est 1.42 1.17 1.39
Noc / All 1.42 1.17 1.39
Noc / Est 1.42 1.17 1.39
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 1

Error D1-bg D1-fg D1-all
All / All 1.91 3.94 2.13
All / Est 1.91 3.94 2.13
Noc / All 1.83 3.94 2.07
Noc / Est 1.83 3.94 2.07
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 2

Error D1-bg D1-fg D1-all
All / All 2.75 10.12 3.11
All / Est 2.75 10.12 3.11
Noc / All 2.56 10.12 2.93
Noc / Est 2.56 10.12 2.93
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 3

Error D1-bg D1-fg D1-all
All / All 2.22 1.38 2.14
All / Est 2.22 1.38 2.14
Noc / All 2.22 1.38 2.14
Noc / Est 2.22 1.38 2.14
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 4

Error D1-bg D1-fg D1-all
All / All 0.88 0.54 0.82
All / Est 0.88 0.54 0.82
Noc / All 0.67 0.54 0.65
Noc / Est 0.67 0.54 0.65
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 5

Error D1-bg D1-fg D1-all
All / All 2.26 2.63 2.29
All / Est 2.26 2.63 2.29
Noc / All 2.19 2.63 2.23
Noc / Est 2.19 2.63 2.23
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 6

Error D1-bg D1-fg D1-all
All / All 2.87 1.03 2.68
All / Est 2.87 1.03 2.68
Noc / All 2.93 1.03 2.73
Noc / Est 2.93 1.03 2.73
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 7

Error D1-bg D1-fg D1-all
All / All 0.36 3.68 1.01
All / Est 0.36 3.68 1.01
Noc / All 0.35 3.68 1.01
Noc / Est 0.35 3.68 1.01
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 8

Error D1-bg D1-fg D1-all
All / All 0.34 1.80 0.61
All / Est 0.34 1.80 0.61
Noc / All 0.33 1.80 0.60
Noc / Est 0.33 1.80 0.60
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 9

Error D1-bg D1-fg D1-all
All / All 0.34 1.23 0.56
All / Est 0.34 1.23 0.56
Noc / All 0.34 1.29 0.57
Noc / Est 0.34 1.29 0.57
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 10

Error D1-bg D1-fg D1-all
All / All 1.10 1.88 1.28
All / Est 1.10 1.88 1.28
Noc / All 1.10 1.88 1.28
Noc / Est 1.10 1.88 1.28
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 11

Error D1-bg D1-fg D1-all
All / All 1.06 0.35 0.93
All / Est 1.06 0.35 0.93
Noc / All 1.07 0.35 0.94
Noc / Est 1.07 0.35 0.94
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 12

Error D1-bg D1-fg D1-all
All / All 0.80 0.77 0.80
All / Est 0.80 0.77 0.80
Noc / All 0.64 0.77 0.65
Noc / Est 0.64 0.77 0.65
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 13

Error D1-bg D1-fg D1-all
All / All 0.51 0.24 0.48
All / Est 0.51 0.24 0.48
Noc / All 0.45 0.24 0.42
Noc / Est 0.45 0.24 0.42
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 14

Error D1-bg D1-fg D1-all
All / All 1.44 0.00 1.42
All / Est 1.44 0.00 1.42
Noc / All 1.28 0.00 1.26
Noc / Est 1.28 0.00 1.26
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 15

Error D1-bg D1-fg D1-all
All / All 2.90 0.39 2.67
All / Est 2.90 0.39 2.67
Noc / All 2.96 0.39 2.72
Noc / Est 2.96 0.39 2.72
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 16

Error D1-bg D1-fg D1-all
All / All 3.37 0.08 2.89
All / Est 3.37 0.08 2.89
Noc / All 3.25 0.08 2.78
Noc / Est 3.25 0.08 2.78
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 17

Error D1-bg D1-fg D1-all
All / All 0.97 0.08 0.88
All / Est 0.97 0.08 0.88
Noc / All 0.97 0.08 0.88
Noc / Est 0.97 0.08 0.88
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 18

Error D1-bg D1-fg D1-all
All / All 4.89 1.08 3.08
All / Est 4.89 1.08 3.08
Noc / All 4.84 1.08 3.04
Noc / Est 4.84 1.08 3.04
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 19

Error D1-bg D1-fg D1-all
All / All 0.78 0.07 0.70
All / Est 0.78 0.07 0.70
Noc / All 0.77 0.07 0.69
Noc / Est 0.77 0.07 0.69
This table as LaTeX

Input Image

D1 Result

D1 Error




eXTReMe Tracker