Method

Towards Attention-based Semantic Aware Guided Depth Completion [SemAttNet]


Submitted on 7 Nov. 2021 13:22 by
Danish Nazir (Technical University of Kaiserslautern)

Running time:0.2 s
Environment:1 core @ 2.5 Ghz (C/C++)

Method Description:
In this paper, we propose a novel three-branch
backbone comprising color-guided, semantic-guided,
and depth-guided branches. Specifically, the
color-guided branch takes a sparse depth map and
RGB image as an input and generates color depth
which includes color cues (e.g., object
boundaries) of the scene. The predicted dense
depth map of color-guided branch along-with
semantic image and sparse depth map is passed as
input to semantic-guided branch for estimating
semantic depth. The depth-guided branch takes
sparse, color, and semantic depths to generate the
dense depth map. The color depth, semantic depth,
and guided depth are adaptively fused to produce
the output of our proposed three-branch backbone.
In addition, we also propose to apply semantic-
aware multi-modal attention-based fusion block
(SAMMAFB) to fuse features between all three
branches. We further use CSPN++ with Atrous
convolutions to refine the dense depth map
produced by our three-branch backbone.
Parameters:
lr=0.0012, bs = 8
Latex Bibtex:
@misc{https://doi.org/10.48550/arxiv.2204.13635,
doi = {10.48550/ARXIV.2204.13635},

url = {https://arxiv.org/abs/2204.13635},

author = {Nazir, Danish and Liwicki, Marcus and
Stricker, Didier and Afzal, Muhammad Zeshan},

keywords = {Computer Vision and Pattern
Recognition (cs.CV), FOS: Computer and information
sciences, FOS: Computer and information sciences},

title = {SemAttNet: Towards Attention-based
Semantic Aware Guided Depth Completion},

publisher = {arXiv},

year = {2022},

copyright = {Creative Commons Attribution Share
Alike 4.0 International}
}

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 2.03 0.90 709.41 205.49
This table as LaTeX

Test Image 0

iRMSE iMAE RMSE MAE
Error 2.96 0.71 843.62 164.62
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 1

iRMSE iMAE RMSE MAE
Error 2.64 0.78 814.78 86.13
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 2

iRMSE iMAE RMSE MAE
Error 2.07 1.46 1134.32 429.22
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 3

iRMSE iMAE RMSE MAE
Error 2.74 1.51 644.35 237.30
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 4

iRMSE iMAE RMSE MAE
Error 2.68 1.40 553.22 214.68
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 5

iRMSE iMAE RMSE MAE
Error 3.53 0.91 858.24 165.23
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 6

iRMSE iMAE RMSE MAE
Error 4.09 1.06 647.48 158.48
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 7

iRMSE iMAE RMSE MAE
Error 3.38 1.19 649.11 164.43
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 8

iRMSE iMAE RMSE MAE
Error 1.93 0.62 667.75 147.64
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 9

iRMSE iMAE RMSE MAE
Error 2.00 1.07 699.90 210.95
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 10

iRMSE iMAE RMSE MAE
Error 1.80 1.28 758.47 393.46
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 11

iRMSE iMAE RMSE MAE
Error 2.47 1.06 1188.24 397.19
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 12

iRMSE iMAE RMSE MAE
Error 3.76 1.54 965.78 249.34
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 13

iRMSE iMAE RMSE MAE
Error 1.11 0.70 579.64 176.51
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 14

iRMSE iMAE RMSE MAE
Error 1.31 0.78 541.92 140.01
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 15

iRMSE iMAE RMSE MAE
Error 3.41 1.28 480.01 161.19
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 16

iRMSE iMAE RMSE MAE
Error 1.33 0.67 572.84 172.45
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 17

iRMSE iMAE RMSE MAE
Error 1.23 0.63 548.64 163.26
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 18

iRMSE iMAE RMSE MAE
Error 1.64 0.80 637.87 238.75
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 19

iRMSE iMAE RMSE MAE
Error 1.04 0.74 647.71 208.60
This table as LaTeX

Input Image

D1 Result

D1 Error




eXTReMe Tracker