Method

Propagating Confidences through CNNs for Sparse Data Regression [NConv-CNN (d)]
https://github.com/abdo-eldesokey/nconv

Submitted on 1 Jun. 2018 16:15 by
Abdelrahman Eldesokey (Linköping University)

Running time:0.01 s
Environment:GPU @ 1.5 Ghz (Python)

Method Description:
[To appear at the British Machine Vision
Conference (BMVC) 2018]
A multi-scale CNN architecture utilizing the
proposed Normalized Convolution layer. The CNN
takes depth data as input and outputs a dense
depth map together with a pixel-wise confidence
measure. Our proposed CNN is efficient (only
480
network parameters) and especially suitable for
different real-world applications where
computational efficiency is desired.
For more details, check the ArXiv submission:
https://arxiv.org/abs/1805.11913
Parameters:
See the paper:
https://arxiv.org/abs/1805.11913
Latex Bibtex:
@article{Eldesokey2018,
archivePrefix = {arXiv},
arxivId = {1805.11913},
author = {Eldesokey, Abdelrahman and Felsberg,
Michael and Khan, Fahad Shahbaz},
eprint = {1805.11913},
month = {may},
title = {{Propagating Confidences through CNNs
for Sparse Data Regression}},
url = {http://arxiv.org/abs/1805.11913},
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 Sparsity Invariant CNNs (THREEDV 2017), 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 Sparsity Invariant CNNs (THREEDV 2017), 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

iRMSE iMAE RMSE MAE
Error 4.67 1.52 1268.22 360.28
This table as LaTeX

Test Image 0

iRMSE iMAE RMSE MAE
Error 5.30 1.45 1540.19 386.57
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 1

iRMSE iMAE RMSE MAE
Error 6.40 1.71 1144.83 220.06
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 2

iRMSE iMAE RMSE MAE
Error 2.73 1.68 1949.73 603.20
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 3

iRMSE iMAE RMSE MAE
Error 5.79 2.36 1096.18 405.27
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 4

iRMSE iMAE RMSE MAE
Error 5.49 2.15 784.75 326.86
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 5

iRMSE iMAE RMSE MAE
Error 6.95 2.01 1468.50 400.21
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 6

iRMSE iMAE RMSE MAE
Error 17.75 4.42 1886.35 450.39
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 7

iRMSE iMAE RMSE MAE
Error 6.43 1.82 1656.46 276.76
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 8

iRMSE iMAE RMSE MAE
Error 7.15 1.61 1195.26 325.41
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 9

iRMSE iMAE RMSE MAE
Error 3.21 1.48 1564.08 417.27
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 10

iRMSE iMAE RMSE MAE
Error 1.94 1.34 1086.93 450.87
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 11

iRMSE iMAE RMSE MAE
Error 4.97 1.82 2106.29 697.04
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 12

iRMSE iMAE RMSE MAE
Error 18.13 6.72 1646.75 676.31
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 13

iRMSE iMAE RMSE MAE
Error 1.73 0.93 1054.69 260.07
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 14

iRMSE iMAE RMSE MAE
Error 7.21 1.81 1004.67 274.10
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 15

iRMSE iMAE RMSE MAE
Error 5.73 2.05 813.65 274.79
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 16

iRMSE iMAE RMSE MAE
Error 2.36 0.99 894.08 293.04
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 17

iRMSE iMAE RMSE MAE
Error 2.25 0.91 898.78 265.69
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 18

iRMSE iMAE RMSE MAE
Error 3.71 1.28 1069.51 377.56
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 19

iRMSE iMAE RMSE MAE
Error 1.75 0.95 1309.46 315.46
This table as LaTeX

Input Image

D1 Result

D1 Error




eXTReMe Tracker