Method

GANet reference submission for robust vision challenge [at] [GANetREF_RVC]
https://github.com/feihuzhang/GANet

Submitted on 17 Jul. 2020 10:22 by
Nicolas Jourdan (TU Darmstadt)

Running time:1.62 s
Environment:GPU @ >3.5 Ghz (Python + C/C++)

Method Description:
GANet reference submission as baseline for the
stereo benchmark of the robust vision challenge
2020.
All method credits go to the original author
(Zhang et al.)

We are formulating traditional geometric and
optimization of stereo into deep neural networks.
Parameters:
Trained on Middleburry, KITTI, ETH3D from the KITTI
checkpoint made available in the GANet repository
on Github by the original authors.
Frequency of sampling was adapted to the dataset
size. Test images scaled to next multiple of 48.
Latex Bibtex:
@inproceedings{Zhang2019GANet,
title={GA-Net: Guided Aggregation Net for End-
to-end Stereo Matching},
author={Zhang, Feihu and Prisacariu, Victor and
Yang, Ruigang and Torr, Philip HS},
booktitle={Proceedings of the IEEE Conference on
Computer Vision and Pattern Recognition},
pages={185--194},
year={2019}
}

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 1.88 4.58 2.33
All / Est 1.88 4.58 2.33
Noc / All 1.74 4.04 2.12
Noc / Est 1.74 4.04 2.12
This table as LaTeX

Test Image 0

Error D1-bg D1-fg D1-all
All / All 2.61 0.51 2.32
All / Est 2.61 0.51 2.32
Noc / All 2.57 0.51 2.28
Noc / Est 2.57 0.51 2.28
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 1

Error D1-bg D1-fg D1-all
All / All 1.68 2.80 1.81
All / Est 1.68 2.80 1.81
Noc / All 1.61 2.80 1.75
Noc / Est 1.61 2.80 1.75
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 2

Error D1-bg D1-fg D1-all
All / All 2.31 2.24 2.30
All / Est 2.31 2.24 2.30
Noc / All 2.07 2.24 2.08
Noc / Est 2.07 2.24 2.08
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 3

Error D1-bg D1-fg D1-all
All / All 2.39 0.28 2.20
All / Est 2.39 0.28 2.20
Noc / All 2.36 0.28 2.17
Noc / Est 2.36 0.28 2.17
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 4

Error D1-bg D1-fg D1-all
All / All 0.90 0.31 0.80
All / Est 0.90 0.31 0.80
Noc / All 0.88 0.31 0.79
Noc / Est 0.88 0.31 0.79
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 5

Error D1-bg D1-fg D1-all
All / All 3.60 1.92 3.45
All / Est 3.60 1.92 3.45
Noc / All 3.22 1.92 3.10
Noc / Est 3.22 1.92 3.10
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 6

Error D1-bg D1-fg D1-all
All / All 4.45 0.96 4.08
All / Est 4.45 0.96 4.08
Noc / All 4.55 0.96 4.16
Noc / Est 4.55 0.96 4.16
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 7

Error D1-bg D1-fg D1-all
All / All 0.96 1.79 1.12
All / Est 0.96 1.79 1.12
Noc / All 0.97 1.79 1.13
Noc / Est 0.97 1.79 1.13
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 8

Error D1-bg D1-fg D1-all
All / All 0.54 2.53 0.91
All / Est 0.54 2.53 0.91
Noc / All 0.54 2.53 0.91
Noc / Est 0.54 2.53 0.91
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 9

Error D1-bg D1-fg D1-all
All / All 0.74 1.73 0.99
All / Est 0.74 1.73 0.99
Noc / All 0.74 1.82 1.01
Noc / Est 0.74 1.82 1.01
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 10

Error D1-bg D1-fg D1-all
All / All 1.75 1.62 1.72
All / Est 1.75 1.62 1.72
Noc / All 1.77 1.62 1.73
Noc / Est 1.77 1.62 1.73
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 11

Error D1-bg D1-fg D1-all
All / All 1.20 0.72 1.11
All / Est 1.20 0.72 1.11
Noc / All 1.15 0.72 1.07
Noc / Est 1.15 0.72 1.07
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 12

Error D1-bg D1-fg D1-all
All / All 0.80 0.79 0.80
All / Est 0.80 0.79 0.80
Noc / All 0.67 0.79 0.68
Noc / Est 0.67 0.79 0.68
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 13

Error D1-bg D1-fg D1-all
All / All 0.99 0.26 0.90
All / Est 0.99 0.26 0.90
Noc / All 0.82 0.26 0.75
Noc / Est 0.82 0.26 0.75
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 14

Error D1-bg D1-fg D1-all
All / All 1.35 0.00 1.33
All / Est 1.35 0.00 1.33
Noc / All 1.21 0.00 1.19
Noc / Est 1.21 0.00 1.19
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 15

Error D1-bg D1-fg D1-all
All / All 3.87 0.24 3.54
All / Est 3.87 0.24 3.54
Noc / All 3.95 0.24 3.61
Noc / Est 3.95 0.24 3.61
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 16

Error D1-bg D1-fg D1-all
All / All 3.50 0.16 3.01
All / Est 3.50 0.16 3.01
Noc / All 3.37 0.16 2.90
Noc / Est 3.37 0.16 2.90
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 17

Error D1-bg D1-fg D1-all
All / All 0.79 0.28 0.74
All / Est 0.79 0.28 0.74
Noc / All 0.78 0.28 0.73
Noc / Est 0.78 0.28 0.73
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 18

Error D1-bg D1-fg D1-all
All / All 5.41 3.34 4.43
All / Est 5.41 3.34 4.43
Noc / All 5.34 3.34 4.38
Noc / Est 5.34 3.34 4.38
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 19

Error D1-bg D1-fg D1-all
All / All 1.11 0.27 1.01
All / Est 1.11 0.27 1.01
Noc / All 1.09 0.27 1.00
Noc / Est 1.09 0.27 1.00
This table as LaTeX

Input Image

D1 Result

D1 Error




eXTReMe Tracker