Method

Unsupervised Learning of Optical Deep Feature Similarity [UnsupSimFlow]
tobeadded

Submitted on 7 Jul. 2020 08:01 by
Woobin Im (Korea Advanced Institute of Science and Technology)

Running time:0.03 s
Environment:8 cores @ 3.0 Ghz (Python + C/C++)

Method Description:
Deep unsupervised learning for optical flow has
been proposed, where the loss
measures image similarity with the warping
function parameterized by estimated
flow. With the fused similarity, our network
better learns flow by minimizing
our proposed feature separation loss. The proposed
method is a polarizing
scheme, resulting in a more discriminative
similarity map.
Parameters:
\lambda_s=1e-4, \lambda_f=2.0
Latex Bibtex:
@InProceedings{Im_2020_ECCV,
author = {Im, Woobin and Kim, Tae-Kyun and Yoon,
Sung-Eui},
title = {Unsupervised Learning of Optical Flow
with Deep Feature Similarity},
booktitle = {The European Conference on Computer
Vision (ECCV)},
month = {August},
year = {2020}
}

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 , 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 , 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 Fl-bg Fl-fg Fl-all
All / All 12.60 17.27 13.38
All / Est 12.60 17.27 13.38
Noc / All 7.06 13.36 8.21
Noc / Est 7.06 13.36 8.21
This table as LaTeX

Test Image 0

Error Fl-bg Fl-fg Fl-all
All / All 5.98 57.00 12.98
All / Est 5.98 57.00 12.98
Noc / All 4.49 57.00 12.45
Noc / Est 4.49 57.00 12.45
This table as LaTeX

Input Image

Flow Result

Flow Error


Test Image 1

Error Fl-bg Fl-fg Fl-all
All / All 5.32 13.77 6.26
All / Est 5.32 13.77 6.26
Noc / All 4.58 13.77 5.72
Noc / Est 4.58 13.77 5.72
This table as LaTeX

Input Image

Flow Result

Flow Error


Test Image 2

Error Fl-bg Fl-fg Fl-all
All / All 10.05 14.55 10.27
All / Est 10.05 14.55 10.27
Noc / All 6.96 14.55 7.41
Noc / Est 6.96 14.55 7.41
This table as LaTeX

Input Image

Flow Result

Flow Error


Test Image 3

Error Fl-bg Fl-fg Fl-all
All / All 18.80 56.41 22.27
All / Est 18.80 56.41 22.27
Noc / All 13.21 41.81 15.68
Noc / Est 13.21 41.81 15.68
This table as LaTeX

Input Image

Flow Result

Flow Error


Test Image 4

Error Fl-bg Fl-fg Fl-all
All / All 16.24 28.34 18.25
All / Est 16.24 28.34 18.25
Noc / All 11.53 25.98 14.23
Noc / Est 11.53 25.98 14.23
This table as LaTeX

Input Image

Flow Result

Flow Error


Test Image 5

Error Fl-bg Fl-fg Fl-all
All / All 12.39 0.00 11.28
All / Est 12.39 0.00 11.28
Noc / All 9.46 0.00 8.49
Noc / Est 9.46 0.00 8.49
This table as LaTeX

Input Image

Flow Result

Flow Error


Test Image 6

Error Fl-bg Fl-fg Fl-all
All / All 10.17 0.75 9.18
All / Est 10.17 0.75 9.18
Noc / All 8.15 0.75 7.27
Noc / Est 8.15 0.75 7.27
This table as LaTeX

Input Image

Flow Result

Flow Error


Test Image 7

Error Fl-bg Fl-fg Fl-all
All / All 2.33 39.34 9.58
All / Est 2.33 39.34 9.58
Noc / All 2.33 40.51 9.47
Noc / Est 2.33 40.51 9.47
This table as LaTeX

Input Image

Flow Result

Flow Error


Test Image 8

Error Fl-bg Fl-fg Fl-all
All / All 2.87 7.60 3.75
All / Est 2.87 7.60 3.75
Noc / All 2.87 7.60 3.75
Noc / Est 2.87 7.60 3.75
This table as LaTeX

Input Image

Flow Result

Flow Error


Test Image 9

Error Fl-bg Fl-fg Fl-all
All / All 2.68 30.20 9.71
All / Est 2.68 30.20 9.71
Noc / All 2.68 30.20 9.71
Noc / Est 2.68 30.20 9.71
This table as LaTeX

Input Image

Flow Result

Flow Error


Test Image 10

Error Fl-bg Fl-fg Fl-all
All / All 7.74 10.68 8.41
All / Est 7.74 10.68 8.41
Noc / All 4.52 10.68 6.12
Noc / Est 4.52 10.68 6.12
This table as LaTeX

Input Image

Flow Result

Flow Error


Test Image 11

Error Fl-bg Fl-fg Fl-all
All / All 5.67 6.18 5.76
All / Est 5.67 6.18 5.76
Noc / All 5.01 6.19 5.25
Noc / Est 5.01 6.19 5.25
This table as LaTeX

Input Image

Flow Result

Flow Error


Test Image 12

Error Fl-bg Fl-fg Fl-all
All / All 5.40 3.86 5.30
All / Est 5.40 3.86 5.30
Noc / All 3.29 3.86 3.33
Noc / Est 3.29 3.86 3.33
This table as LaTeX

Input Image

Flow Result

Flow Error


Test Image 13

Error Fl-bg Fl-fg Fl-all
All / All 6.33 15.20 7.42
All / Est 6.33 15.20 7.42
Noc / All 2.49 8.61 3.20
Noc / Est 2.49 8.61 3.20
This table as LaTeX

Input Image

Flow Result

Flow Error


Test Image 14

Error Fl-bg Fl-fg Fl-all
All / All 5.76 7.81 5.80
All / Est 5.76 7.81 5.80
Noc / All 2.90 7.81 3.00
Noc / Est 2.90 7.81 3.00
This table as LaTeX

Input Image

Flow Result

Flow Error


Test Image 15

Error Fl-bg Fl-fg Fl-all
All / All 17.32 9.02 16.57
All / Est 17.32 9.02 16.57
Noc / All 9.18 9.02 9.16
Noc / Est 9.18 9.02 9.16
This table as LaTeX

Input Image

Flow Result

Flow Error


Test Image 16

Error Fl-bg Fl-fg Fl-all
All / All 23.66 15.60 22.47
All / Est 23.66 15.60 22.47
Noc / All 13.60 15.60 13.96
Noc / Est 13.60 15.60 13.96
This table as LaTeX

Input Image

Flow Result

Flow Error


Test Image 17

Error Fl-bg Fl-fg Fl-all
All / All 9.94 20.42 11.03
All / Est 9.94 20.42 11.03
Noc / All 5.88 20.42 7.76
Noc / Est 5.88 20.42 7.76
This table as LaTeX

Input Image

Flow Result

Flow Error


Test Image 18

Error Fl-bg Fl-fg Fl-all
All / All 25.04 97.17 59.31
All / Est 25.04 97.17 59.31
Noc / All 13.48 90.98 34.85
Noc / Est 13.48 90.98 34.85
This table as LaTeX

Input Image

Flow Result

Flow Error


Test Image 19

Error Fl-bg Fl-fg Fl-all
All / All 17.77 11.81 17.10
All / Est 17.77 11.81 17.10
Noc / All 7.63 11.81 8.24
Noc / Est 7.63 11.81 8.24
This table as LaTeX

Input Image

Flow Result

Flow Error




eXTReMe Tracker