Method

Non-Local Affinity Distillation Network for Lightweight Depth Completion [ADNet_Small]


Submitted on 18 Sep. 2023 09:55 by
Jang Hyun Kim (Pusan national University)

Running time:0.05 s
Environment:1 core @ 2.5 Ghz (Python)

Method Description:
In these real-world applications, it is strictly required for a depth completion model to achieve lightweight network architectures and reduced computational costs to operate in real-time. With this in mind, we propose a non-local affinity distillation network for lightweight depth completion. Our method utilizes a minimal number of neighbors with strong affinities for fast spatial propagation by optimal transport between teacher and student affinities through knowledge distillation. Moreover, we propose to utilize guidance from missing LiDAR points to further improve the performance of our network especially in regions without LiDAR points.
Parameters:
Please refer to the paper.
Latex Bibtex:
@article{kim2024adnet,
title={ADNet: Non-Local Affinity Distillation Network for Lightweight Depth Completion With Guidance From Missing LiDAR Points},
author={Kim, Janghyun and Noh, Jeonghyun and Jeong, Mingyu and Lee, Wonju and Park, Yeonchool and Park, Jinsun},
journal={IEEE Robotics and Automation Letters},
year={2024},
publisher={IEEE}
}

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 2.07 0.88 767.17 209.44
This table as LaTeX

Test Image 0

iRMSE iMAE RMSE MAE
Error 3.31 0.73 883.31 170.39
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 1

iRMSE iMAE RMSE MAE
Error 2.48 0.71 715.91 78.05
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 2

iRMSE iMAE RMSE MAE
Error 2.08 1.41 1198.16 429.89
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 3

iRMSE iMAE RMSE MAE
Error 2.76 1.48 638.64 235.80
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 4

iRMSE iMAE RMSE MAE
Error 2.56 1.37 530.43 214.56
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 5

iRMSE iMAE RMSE MAE
Error 3.14 0.88 745.72 163.88
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 6

iRMSE iMAE RMSE MAE
Error 3.21 0.94 447.66 154.16
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 7

iRMSE iMAE RMSE MAE
Error 3.53 1.14 719.04 170.69
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 8

iRMSE iMAE RMSE MAE
Error 1.80 0.65 804.26 162.96
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 9

iRMSE iMAE RMSE MAE
Error 1.88 1.03 765.82 212.08
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 10

iRMSE iMAE RMSE MAE
Error 1.73 1.21 744.99 375.11
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 11

iRMSE iMAE RMSE MAE
Error 2.45 1.05 1183.40 386.03
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 12

iRMSE iMAE RMSE MAE
Error 3.57 1.38 959.37 239.78
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 13

iRMSE iMAE RMSE MAE
Error 1.24 0.73 668.98 188.04
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 14

iRMSE iMAE RMSE MAE
Error 1.27 0.72 558.91 138.35
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 15

iRMSE iMAE RMSE MAE
Error 3.08 1.19 517.10 161.72
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 16

iRMSE iMAE RMSE MAE
Error 1.35 0.65 592.93 172.18
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 17

iRMSE iMAE RMSE MAE
Error 1.32 0.63 603.93 165.85
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 18

iRMSE iMAE RMSE MAE
Error 1.64 0.87 649.48 248.14
This table as LaTeX

Input Image

D1 Result

D1 Error


Test Image 19

iRMSE iMAE RMSE MAE
Error 1.05 0.74 681.46 213.15
This table as LaTeX

Input Image

D1 Result

D1 Error




eXTReMe Tracker