Method

Cost Affinity Learning Network for Stereo Matching [CAL-Net]


Submitted on 31 Oct. 2020 04:13 by
Hong Zhang (Dalian University of Technology)

Running time:0.44 s
Environment:2 cores @ 2.5 Ghz (Python)

Method Description:
For stereo matching problem, most existing methods
directlyaggregate Convolutional Neural
Network(CNN) output fea-tures to obtain more
discriminative cost features, but ignorethe
affinity of each element in cost feature which
also playsa key role in enhancing the cost
feature. In this work, wepropose a novel cost
affinity learning network.
Parameters:
n=4,\gamma = 1
Latex Bibtex:
@inproceedings{DBLP:conf/icassp/ChenLWZLW21,
author = {Shenglun Chen and
Baopu Li and
Wei Wang and
Hong Zhang and
Haojie Li and
Zhihui Wang},
title = {Cost Affinity Learning Network for
Stereo Matching},
booktitle = {{IEEE} International Conference on
Acoustics, Speech and Signal Processing,
{ICASSP} 2021, Toronto, ON, Canada,
June 6-11, 2021},
pages = {2120--2124},
publisher = {{IEEE}},
year = {2021},
url =
{https://doi.org/10.1109/ICASSP39728.2021.9413556}
,
doi = {10.1109/ICASSP39728.2021.9413556},
timestamp = {Thu, 08 Jul 2021 17:13:36 +0200},
biburl =
{https://dblp.org/rec/conf/icassp/ChenLWZLW21.bib}
,
bibsource = {dblp computer science bibliography,
https://dblp.org}
}

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 1.59 3.76 1.95
All / Est 1.59 3.76 1.95
Noc / All 1.45 3.42 1.77
Noc / Est 1.45 3.42 1.77
This table as LaTeX

Test Image 0

Error D1-bg D1-fg D1-all
All / All 1.52 0.81 1.43
All / Est 1.52 0.81 1.43
Noc / All 1.53 0.81 1.43
Noc / Est 1.53 0.81 1.43
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 1

Error D1-bg D1-fg D1-all
All / All 1.56 3.64 1.80
All / Est 1.56 3.64 1.80
Noc / All 1.49 3.64 1.73
Noc / Est 1.49 3.64 1.73
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 2

Error D1-bg D1-fg D1-all
All / All 2.06 1.28 2.02
All / Est 2.06 1.28 2.02
Noc / All 1.85 1.28 1.83
Noc / Est 1.85 1.28 1.83
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 3

Error D1-bg D1-fg D1-all
All / All 1.97 1.92 1.96
All / Est 1.97 1.92 1.96
Noc / All 1.92 1.92 1.92
Noc / Est 1.92 1.92 1.92
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 4

Error D1-bg D1-fg D1-all
All / All 1.18 0.33 1.04
All / Est 1.18 0.33 1.04
Noc / All 1.19 0.33 1.05
Noc / Est 1.19 0.33 1.05
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 5

Error D1-bg D1-fg D1-all
All / All 2.81 0.97 2.64
All / Est 2.81 0.97 2.64
Noc / All 2.60 0.97 2.45
Noc / Est 2.60 0.97 2.45
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 6

Error D1-bg D1-fg D1-all
All / All 3.98 0.42 3.61
All / Est 3.98 0.42 3.61
Noc / All 4.07 0.42 3.68
Noc / Est 4.07 0.42 3.68
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 7

Error D1-bg D1-fg D1-all
All / All 0.51 2.13 0.82
All / Est 0.51 2.13 0.82
Noc / All 0.51 2.13 0.83
Noc / Est 0.51 2.13 0.83
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 8

Error D1-bg D1-fg D1-all
All / All 0.46 1.53 0.65
All / Est 0.46 1.53 0.65
Noc / All 0.45 1.53 0.65
Noc / Est 0.45 1.53 0.65
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 9

Error D1-bg D1-fg D1-all
All / All 0.44 0.61 0.48
All / Est 0.44 0.61 0.48
Noc / All 0.45 0.64 0.50
Noc / Est 0.45 0.64 0.50
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 10

Error D1-bg D1-fg D1-all
All / All 1.14 2.61 1.48
All / Est 1.14 2.61 1.48
Noc / All 1.14 2.61 1.48
Noc / Est 1.14 2.61 1.48
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 11

Error D1-bg D1-fg D1-all
All / All 1.13 0.88 1.09
All / Est 1.13 0.88 1.09
Noc / All 1.14 0.88 1.09
Noc / Est 1.14 0.88 1.09
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 12

Error D1-bg D1-fg D1-all
All / All 0.72 0.82 0.73
All / Est 0.72 0.82 0.73
Noc / All 0.58 0.82 0.60
Noc / Est 0.58 0.82 0.60
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 13

Error D1-bg D1-fg D1-all
All / All 0.65 0.16 0.59
All / Est 0.65 0.16 0.59
Noc / All 0.60 0.16 0.54
Noc / Est 0.60 0.16 0.54
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 14

Error D1-bg D1-fg D1-all
All / All 1.42 0.11 1.39
All / Est 1.42 0.11 1.39
Noc / All 1.30 0.11 1.28
Noc / Est 1.30 0.11 1.28
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 15

Error D1-bg D1-fg D1-all
All / All 2.93 0.04 2.67
All / Est 2.93 0.04 2.67
Noc / All 2.99 0.04 2.71
Noc / Est 2.99 0.04 2.71
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 16

Error D1-bg D1-fg D1-all
All / All 4.10 0.17 3.52
All / Est 4.10 0.17 3.52
Noc / All 3.94 0.17 3.38
Noc / Est 3.94 0.17 3.38
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 17

Error D1-bg D1-fg D1-all
All / All 0.97 0.04 0.88
All / Est 0.97 0.04 0.88
Noc / All 0.97 0.04 0.87
Noc / Est 0.97 0.04 0.87
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 18

Error D1-bg D1-fg D1-all
All / All 4.81 0.64 2.82
All / Est 4.81 0.64 2.82
Noc / All 4.73 0.64 2.77
Noc / Est 4.73 0.64 2.77
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 19

Error D1-bg D1-fg D1-all
All / All 0.83 1.20 0.87
All / Est 0.83 1.20 0.87
Noc / All 0.81 1.20 0.85
Noc / Est 0.81 1.20 0.85
This table as LaTeX

Input Image

D1 Result

D1 Error




eXTReMe Tracker