Method

RecResNet: A Recurrent Residual CNN Architecture for Disparity Map Enhancement [RecResNet]
https://github.com/kbatsos/RecResNet

Submitted on 1 Mar. 2019 19:30 by
KONSTANTINOS BATSOS (Stevens Institute of technology)

Running time:0.3 s
Environment:GPU @ NVIDIA TITAN X (Tensorflow)

Method Description:
We present a neural network architecture applied to
the
problem of refining a dense disparity map generated
by
a stereo algorithm to which we have no access. Our
approach is able to learn which disparity values
should be
modified and how, from a training set of images,
estimated
disparity maps and the corresponding ground truth.
Its only
input at test time is a disparity map and the
reference image. Two design characteristics are
critical for the success
of our network: (i) it is formulated as a recurrent
neural
network, and (ii) it estimates the output refined
disparity
map as a combination of residuals computed at
multiple
scales, that is at different up-sampling and down-
sampling
rates. The first property allows the network, which
we named
RecResNet, to progressively improve the disparity
map, while
the second property allows the corrections to come
from
different scales of analysis, addressing different
types of
errors in the current disparity map. We present
competitive quantitative and qualitative results on
the KITTI 2012
and 2015 benchmarks that surpass the accuracy of
previous disparity refinement methods.
Parameters:
See our paper
Latex Bibtex:
@inproceedings{batsos2018recresnet,
title={RecResNet: A Recurrent Residual CNN
Architecture for Disparity Map Enhancement},
author={Batsos, Konstantinos and Mordohai,
Philipos},
booktitle={ In International Conference on 3D
Vision (3DV) },
year={2018}
}

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 2.46 6.30 3.10
All / Est 2.46 6.30 3.10
Noc / All 2.23 5.37 2.75
Noc / Est 2.23 5.37 2.75
This table as LaTeX

Test Image 0

Error D1-bg D1-fg D1-all
All / All 2.88 0.87 2.60
All / Est 2.88 0.87 2.60
Noc / All 2.84 0.87 2.57
Noc / Est 2.84 0.87 2.57
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 1

Error D1-bg D1-fg D1-all
All / All 2.12 3.19 2.24
All / Est 2.12 3.19 2.24
Noc / All 2.04 3.19 2.17
Noc / Est 2.04 3.19 2.17
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 2

Error D1-bg D1-fg D1-all
All / All 3.77 4.58 3.81
All / Est 3.77 4.58 3.81
Noc / All 3.49 4.58 3.55
Noc / Est 3.49 4.58 3.55
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 3

Error D1-bg D1-fg D1-all
All / All 3.33 4.75 3.46
All / Est 3.33 4.75 3.46
Noc / All 3.06 4.75 3.22
Noc / Est 3.06 4.75 3.22
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 4

Error D1-bg D1-fg D1-all
All / All 3.79 1.80 3.46
All / Est 3.79 1.80 3.46
Noc / All 3.38 1.80 3.12
Noc / Est 3.38 1.80 3.12
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 5

Error D1-bg D1-fg D1-all
All / All 5.42 1.53 5.07
All / Est 5.42 1.53 5.07
Noc / All 4.55 1.53 4.27
Noc / Est 4.55 1.53 4.27
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 6

Error D1-bg D1-fg D1-all
All / All 7.02 0.85 6.37
All / Est 7.02 0.85 6.37
Noc / All 7.17 0.85 6.49
Noc / Est 7.17 0.85 6.49
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 7

Error D1-bg D1-fg D1-all
All / All 0.79 5.15 1.65
All / Est 0.79 5.15 1.65
Noc / All 0.81 5.15 1.67
Noc / Est 0.81 5.15 1.67
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 8

Error D1-bg D1-fg D1-all
All / All 0.87 1.20 0.94
All / Est 0.87 1.20 0.94
Noc / All 0.87 1.20 0.93
Noc / Est 0.87 1.20 0.93
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 9

Error D1-bg D1-fg D1-all
All / All 1.02 1.39 1.12
All / Est 1.02 1.39 1.12
Noc / All 1.03 1.46 1.14
Noc / Est 1.03 1.46 1.14
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 10

Error D1-bg D1-fg D1-all
All / All 1.40 2.65 1.69
All / Est 1.40 2.65 1.69
Noc / All 1.40 2.65 1.68
Noc / Est 1.40 2.65 1.68
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 11

Error D1-bg D1-fg D1-all
All / All 0.93 0.27 0.81
All / Est 0.93 0.27 0.81
Noc / All 0.93 0.27 0.81
Noc / Est 0.93 0.27 0.81
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 12

Error D1-bg D1-fg D1-all
All / All 0.54 0.55 0.54
All / Est 0.54 0.55 0.54
Noc / All 0.54 0.55 0.54
Noc / Est 0.54 0.55 0.54
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 13

Error D1-bg D1-fg D1-all
All / All 0.74 0.23 0.67
All / Est 0.74 0.23 0.67
Noc / All 0.58 0.23 0.53
Noc / Est 0.58 0.23 0.53
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 14

Error D1-bg D1-fg D1-all
All / All 1.28 0.22 1.27
All / Est 1.28 0.22 1.27
Noc / All 1.19 0.22 1.17
Noc / Est 1.19 0.22 1.17
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 15

Error D1-bg D1-fg D1-all
All / All 2.73 0.07 2.49
All / Est 2.73 0.07 2.49
Noc / All 2.79 0.07 2.54
Noc / Est 2.79 0.07 2.54
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 16

Error D1-bg D1-fg D1-all
All / All 3.68 0.81 3.26
All / Est 3.68 0.81 3.26
Noc / All 3.51 0.81 3.11
Noc / Est 3.51 0.81 3.11
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 17

Error D1-bg D1-fg D1-all
All / All 0.82 0.08 0.74
All / Est 0.82 0.08 0.74
Noc / All 0.76 0.08 0.69
Noc / Est 0.76 0.08 0.69
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 18

Error D1-bg D1-fg D1-all
All / All 5.11 3.38 4.29
All / Est 5.11 3.38 4.29
Noc / All 5.06 3.38 4.25
Noc / Est 5.06 3.38 4.25
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 19

Error D1-bg D1-fg D1-all
All / All 0.92 0.08 0.82
All / Est 0.92 0.08 0.82
Noc / All 0.89 0.08 0.80
Noc / Est 0.89 0.08 0.80
This table as LaTeX

Input Image

D1 Result

D1 Error




eXTReMe Tracker