Method

SemARFlow [at] [SemARFlow]
https://github.com/duke-vision/semantic-unsup-flow-release

Submitted on 21 Feb. 2023 04:11 by
Shuai Yuan (Duke University)

Running time:0.0168s
Environment:GPU @ 2.5 Ghz (Python)

Method Description:
We introduce SemARFlow, an unsupervised optical flow
network designed for autonomous driving data that
takes estimated semantic segmentation masks as
additional inputs. This additional information is
injected into the encoder and into a learned
upsampler that refines the flow output. In addition,
a simple yet effective semantic augmentation module
provides self-supervision when learning flow and its
boundaries for vehicles, poles, and sky.
Parameters:
See paper.
Latex Bibtex:
@InProceedings{yuan2023semarflow,
author = {Yuan, Shuai and Yu, Shuzhi and Kim,
Hannah and Tomasi, Carlo},
title = {SemARFlow: Injecting Semantics into
Unsupervised Optical Flow Estimation for Autonomous
Driving},
booktitle = ICCV,
month = {October},
year = {2023},
pages = {9566-9577}
}

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 Fl-bg Fl-fg Fl-all
All / All 7.48 12.91 8.38
All / Est 7.48 12.91 8.38
Noc / All 4.58 9.30 5.43
Noc / Est 4.58 9.30 5.43
This table as LaTeX

Test Image 0

Error Fl-bg Fl-fg Fl-all
All / All 2.51 7.87 3.25
All / Est 2.51 7.87 3.25
Noc / All 2.57 7.87 3.38
Noc / Est 2.57 7.87 3.38
This table as LaTeX

Input Image

Flow Result

Flow Error


Test Image 1

Error Fl-bg Fl-fg Fl-all
All / All 3.09 13.24 4.22
All / Est 3.09 13.24 4.22
Noc / All 2.92 13.24 4.19
Noc / Est 2.92 13.24 4.19
This table as LaTeX

Input Image

Flow Result

Flow Error


Test Image 2

Error Fl-bg Fl-fg Fl-all
All / All 7.88 15.48 8.25
All / Est 7.88 15.48 8.25
Noc / All 4.92 15.48 5.53
Noc / Est 4.92 15.48 5.53
This table as LaTeX

Input Image

Flow Result

Flow Error


Test Image 3

Error Fl-bg Fl-fg Fl-all
All / All 11.66 55.43 15.70
All / Est 11.66 55.43 15.70
Noc / All 8.64 40.50 11.38
Noc / Est 8.64 40.50 11.38
This table as LaTeX

Input Image

Flow Result

Flow Error


Test Image 4

Error Fl-bg Fl-fg Fl-all
All / All 5.39 17.47 7.39
All / Est 5.39 17.47 7.39
Noc / All 4.69 17.54 7.09
Noc / Est 4.69 17.54 7.09
This table as LaTeX

Input Image

Flow Result

Flow Error


Test Image 5

Error Fl-bg Fl-fg Fl-all
All / All 6.65 0.00 6.05
All / Est 6.65 0.00 6.05
Noc / All 5.00 0.00 4.49
Noc / Est 5.00 0.00 4.49
This table as LaTeX

Input Image

Flow Result

Flow Error


Test Image 6

Error Fl-bg Fl-fg Fl-all
All / All 6.83 0.13 6.13
All / Est 6.83 0.13 6.13
Noc / All 5.11 0.13 4.52
Noc / Est 5.11 0.13 4.52
This table as LaTeX

Input Image

Flow Result

Flow Error


Test Image 7

Error Fl-bg Fl-fg Fl-all
All / All 1.70 24.84 6.23
All / Est 1.70 24.84 6.23
Noc / All 1.70 22.88 5.66
Noc / Est 1.70 22.88 5.66
This table as LaTeX

Input Image

Flow Result

Flow Error


Test Image 8

Error Fl-bg Fl-fg Fl-all
All / All 0.74 9.14 2.29
All / Est 0.74 9.14 2.29
Noc / All 0.74 9.14 2.29
Noc / Est 0.74 9.14 2.29
This table as LaTeX

Input Image

Flow Result

Flow Error


Test Image 9

Error Fl-bg Fl-fg Fl-all
All / All 0.85 16.60 4.88
All / Est 0.85 16.60 4.88
Noc / All 0.85 16.60 4.88
Noc / Est 0.85 16.60 4.88
This table as LaTeX

Input Image

Flow Result

Flow Error


Test Image 10

Error Fl-bg Fl-fg Fl-all
All / All 3.48 9.95 4.96
All / Est 3.48 9.95 4.96
Noc / All 3.35 9.95 5.06
Noc / Est 3.35 9.95 5.06
This table as LaTeX

Input Image

Flow Result

Flow Error


Test Image 11

Error Fl-bg Fl-fg Fl-all
All / All 4.18 4.33 4.21
All / Est 4.18 4.33 4.21
Noc / All 4.30 4.33 4.31
Noc / Est 4.30 4.33 4.31
This table as LaTeX

Input Image

Flow Result

Flow Error


Test Image 12

Error Fl-bg Fl-fg Fl-all
All / All 5.03 3.75 4.94
All / Est 5.03 3.75 4.94
Noc / All 2.66 3.75 2.75
Noc / Est 2.66 3.75 2.75
This table as LaTeX

Input Image

Flow Result

Flow Error


Test Image 13

Error Fl-bg Fl-fg Fl-all
All / All 4.90 8.30 5.32
All / Est 4.90 8.30 5.32
Noc / All 2.18 7.47 2.80
Noc / Est 2.18 7.47 2.80
This table as LaTeX

Input Image

Flow Result

Flow Error


Test Image 14

Error Fl-bg Fl-fg Fl-all
All / All 4.79 3.88 4.77
All / Est 4.79 3.88 4.77
Noc / All 2.68 3.88 2.70
Noc / Est 2.68 3.88 2.70
This table as LaTeX

Input Image

Flow Result

Flow Error


Test Image 15

Error Fl-bg Fl-fg Fl-all
All / All 11.65 4.46 11.00
All / Est 11.65 4.46 11.00
Noc / All 8.75 4.46 8.26
Noc / Est 8.75 4.46 8.26
This table as LaTeX

Input Image

Flow Result

Flow Error


Test Image 16

Error Fl-bg Fl-fg Fl-all
All / All 17.87 24.42 18.84
All / Est 17.87 24.42 18.84
Noc / All 10.45 24.42 12.96
Noc / Est 10.45 24.42 12.96
This table as LaTeX

Input Image

Flow Result

Flow Error


Test Image 17

Error Fl-bg Fl-fg Fl-all
All / All 8.18 16.29 9.03
All / Est 8.18 16.29 9.03
Noc / All 5.81 16.29 7.17
Noc / Est 5.81 16.29 7.17
This table as LaTeX

Input Image

Flow Result

Flow Error


Test Image 18

Error Fl-bg Fl-fg Fl-all
All / All 21.04 72.29 45.39
All / Est 21.04 72.29 45.39
Noc / All 12.42 23.35 15.43
Noc / Est 12.42 23.35 15.43
This table as LaTeX

Input Image

Flow Result

Flow Error


Test Image 19

Error Fl-bg Fl-fg Fl-all
All / All 6.51 12.58 7.20
All / Est 6.51 12.58 7.20
Noc / All 4.67 12.58 5.81
Noc / Est 4.67 12.58 5.81
This table as LaTeX

Input Image

Flow Result

Flow Error




eXTReMe Tracker