Method

NVStereoNet: An Efficient Semi-Supervised Deep Neural Network Approach [NVStereoNet]
https://github.com/NVIDIA-Jetson/redtail/tree/master/stereoDNN

Submitted on 24 May. 2018 21:06 by
Alexey Kamenev (NVIDIA)

Running time:0.6 s
Environment:NVIDIA Titan Xp

Method Description:
We propose a novel semi-supervised learning
approach to training a deep stereo neural
network, along with a novel architecture
containing a machine-learned argmax layer and a
custom, publicly available runtime that enables a
smaller version of our stereo DNN to run on an
embedded GPU.
Parameters:
TBA
Latex Bibtex:
@article{smolyanskiy2018nvstereo,
title={On the Importance of Stereo for Accurate
Depth Estimation: An Efficient Semi-Supervised
Deep Neural Network Approach},
author={Smolyanskiy, Nikolai and Kamenev, Alexey
and Birchfield, Stan},
url={https://arxiv.org/abs/1803.09719},
journal={arXiv preprint arXiv:1803.09719},
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 2.62 5.69 3.13
All / Est 2.62 5.69 3.13
Noc / All 2.03 4.41 2.42
Noc / Est 2.03 4.41 2.42
This table as LaTeX

Test Image 0

Error D1-bg D1-fg D1-all
All / All 2.22 1.89 2.17
All / Est 2.22 1.89 2.17
Noc / All 1.81 1.89 1.82
Noc / Est 1.81 1.89 1.82
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 1

Error D1-bg D1-fg D1-all
All / All 2.67 3.61 2.77
All / Est 2.66 3.61 2.77
Noc / All 2.23 3.61 2.39
Noc / Est 2.23 3.61 2.39
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 2

Error D1-bg D1-fg D1-all
All / All 3.73 0.98 3.59
All / Est 3.73 0.98 3.59
Noc / All 3.20 0.98 3.09
Noc / Est 3.19 0.98 3.08
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 3

Error D1-bg D1-fg D1-all
All / All 3.51 0.71 3.25
All / Est 3.51 0.71 3.25
Noc / All 3.18 0.71 2.95
Noc / Est 3.18 0.71 2.95
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 4

Error D1-bg D1-fg D1-all
All / All 2.29 1.65 2.18
All / Est 2.29 1.65 2.18
Noc / All 1.67 1.65 1.67
Noc / Est 1.67 1.65 1.67
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 5

Error D1-bg D1-fg D1-all
All / All 4.53 2.30 4.33
All / Est 4.53 2.28 4.33
Noc / All 3.45 2.30 3.35
Noc / Est 3.45 2.28 3.34
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 6

Error D1-bg D1-fg D1-all
All / All 6.09 1.12 5.57
All / Est 6.09 1.12 5.57
Noc / All 5.60 1.12 5.12
Noc / Est 5.60 1.12 5.12
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 7

Error D1-bg D1-fg D1-all
All / All 1.94 3.45 2.23
All / Est 1.94 3.45 2.23
Noc / All 1.34 3.45 1.76
Noc / Est 1.34 3.45 1.76
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 8

Error D1-bg D1-fg D1-all
All / All 1.29 2.47 1.51
All / Est 1.29 2.47 1.51
Noc / All 1.25 2.47 1.48
Noc / Est 1.25 2.47 1.48
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 9

Error D1-bg D1-fg D1-all
All / All 2.22 2.12 2.20
All / Est 2.22 2.12 2.19
Noc / All 1.81 1.17 1.65
Noc / Est 1.81 1.17 1.65
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 10

Error D1-bg D1-fg D1-all
All / All 1.82 4.30 2.39
All / Est 1.82 4.30 2.39
Noc / All 1.51 4.30 2.15
Noc / Est 1.51 4.30 2.15
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 11

Error D1-bg D1-fg D1-all
All / All 1.65 1.22 1.57
All / Est 1.65 1.22 1.57
Noc / All 1.38 1.22 1.35
Noc / Est 1.38 1.22 1.35
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 12

Error D1-bg D1-fg D1-all
All / All 1.13 1.53 1.16
All / Est 1.13 1.53 1.16
Noc / All 0.73 1.53 0.78
Noc / Est 0.73 1.53 0.78
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 13

Error D1-bg D1-fg D1-all
All / All 1.20 0.14 1.07
All / Est 1.20 0.14 1.07
Noc / All 0.77 0.14 0.69
Noc / Est 0.77 0.14 0.69
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 14

Error D1-bg D1-fg D1-all
All / All 1.73 0.79 1.72
All / Est 1.73 0.79 1.71
Noc / All 1.38 0.79 1.37
Noc / Est 1.38 0.79 1.37
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 15

Error D1-bg D1-fg D1-all
All / All 3.65 0.12 3.33
All / Est 3.65 0.12 3.33
Noc / All 3.29 0.12 3.00
Noc / Est 3.29 0.12 3.00
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 16

Error D1-bg D1-fg D1-all
All / All 5.28 0.10 4.51
All / Est 5.27 0.10 4.51
Noc / All 4.90 0.10 4.19
Noc / Est 4.90 0.10 4.19
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 17

Error D1-bg D1-fg D1-all
All / All 1.71 0.24 1.55
All / Est 1.71 0.24 1.55
Noc / All 1.37 0.24 1.25
Noc / Est 1.37 0.24 1.25
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 18

Error D1-bg D1-fg D1-all
All / All 5.70 0.84 3.39
All / Est 5.70 0.84 3.39
Noc / All 4.98 0.84 3.00
Noc / Est 4.98 0.84 3.00
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 19

Error D1-bg D1-fg D1-all
All / All 1.39 1.90 1.45
All / Est 1.39 1.90 1.45
Noc / All 1.00 1.90 1.10
Noc / Est 1.00 1.90 1.10
This table as LaTeX

Input Image

D1 Result

D1 Error




eXTReMe Tracker