Self-Supervised Convolutional Neural Networks for Plant Reconstruction Using Stereo Imagery

Yuanxin Xia, Pablo d'Angelo, Jiaojiao Tian, Friedrich Fraundorfer, Peter Reinartz

Research output: Contribution to journal › Article › Research › peer-review

Abstract

Stereo matching can provide complete and dense three-dimensional reconstruction to study plant growth. Recently, high-quality stereo matching results were achieved combining Semi-Global Matching (SGM) with deep learning. However, due to a lack of suitable training data, this technique is not readily applicable for plant reconstruction. We propose a self-supervised Matching Cost with a Convolutional Neural Network (MC-CNN) scheme to calculate matching cost and test it for plant reconstruction. The MC-CNN network is retrained using the initial matching results obtained from the standard MC-CNN weights. For the experiment, closerange photogrammetric imagery of an in-house plant is used. The results show that the performance of self-supervised MC-CNN is superior to the Census algorithm and comparable to MC-CNN trained by a Light Detection and Ranging point cloud. Another experiment is performed …
Original languageEnglish
Pages (from-to)389-399
Number of pages11
JournalPhotogrammetric engineering & remote sensing
Volume85
Issue number5
DOIs
Publication statusPublished - 2019

Fingerprint

imagery
Neural networks
cost
Costs
census
experiment
learning
Experiments

Cite this

Self-Supervised Convolutional Neural Networks for Plant Reconstruction Using Stereo Imagery. / Xia, Yuanxin; d'Angelo, Pablo; Tian, Jiaojiao; Fraundorfer, Friedrich; Reinartz, Peter.

In: Photogrammetric engineering & remote sensing, Vol. 85, No. 5, 2019, p. 389-399.

Research output: Contribution to journal › Article › Research › peer-review

Xia, Y, d'Angelo, P, Tian, J, Fraundorfer, F & Reinartz, P 2019, 'Self-Supervised Convolutional Neural Networks for Plant Reconstruction Using Stereo Imagery' Photogrammetric engineering & remote sensing, vol. 85, no. 5, pp. 389-399. https://doi.org/10.14358/PERS.85.5.389
Xia Y, d'Angelo P, Tian J, Fraundorfer F, Reinartz P. Self-Supervised Convolutional Neural Networks for Plant Reconstruction Using Stereo Imagery. Photogrammetric engineering & remote sensing. 2019;85(5):389-399. https://doi.org/10.14358/PERS.85.5.389
Xia, Yuanxin ; d'Angelo, Pablo ; Tian, Jiaojiao ; Fraundorfer, Friedrich ; Reinartz, Peter. / Self-Supervised Convolutional Neural Networks for Plant Reconstruction Using Stereo Imagery. In: Photogrammetric engineering & remote sensing. 2019 ; Vol. 85, No. 5. pp. 389-399.
@article{758a0aa7acf34ac29f21782d5a6409fe,
title = "Self-Supervised Convolutional Neural Networks for Plant Reconstruction Using Stereo Imagery",
abstract = "Stereo matching can provide complete and dense three-dimensional reconstruction to study plant growth. Recently, high-quality stereo matching results were achieved combining Semi-Global Matching (SGM) with deep learning. However, due to a lack of suitable training data, this technique is not readily applicable for plant reconstruction. We propose a self-supervised Matching Cost with a Convolutional Neural Network (MC-CNN) scheme to calculate matching cost and test it for plant reconstruction. The MC-CNN network is retrained using the initial matching results obtained from the standard MC-CNN weights. For the experiment, closerange photogrammetric imagery of an in-house plant is used. The results show that the performance of self-supervised MC-CNN is superior to the Census algorithm and comparable to MC-CNN trained by a Light Detection and Ranging point cloud. Another experiment is performed …",
author = "Yuanxin Xia and Pablo d'Angelo and Jiaojiao Tian and Friedrich Fraundorfer and Peter Reinartz",
year = "2019",
doi = "10.14358/PERS.85.5.389",
language = "English",
volume = "85",
pages = "389--399",
journal = "Photogrammetric engineering & remote sensing",
issn = "0099-1112",
publisher = "American Society for Photogrammetry and Remote Sensing",
number = "5",

}

TY - JOUR

T1 - Self-Supervised Convolutional Neural Networks for Plant Reconstruction Using Stereo Imagery

AU - Xia, Yuanxin

AU - d'Angelo, Pablo

AU - Tian, Jiaojiao

AU - Fraundorfer, Friedrich

AU - Reinartz, Peter

PY - 2019

Y1 - 2019

N2 - Stereo matching can provide complete and dense three-dimensional reconstruction to study plant growth. Recently, high-quality stereo matching results were achieved combining Semi-Global Matching (SGM) with deep learning. However, due to a lack of suitable training data, this technique is not readily applicable for plant reconstruction. We propose a self-supervised Matching Cost with a Convolutional Neural Network (MC-CNN) scheme to calculate matching cost and test it for plant reconstruction. The MC-CNN network is retrained using the initial matching results obtained from the standard MC-CNN weights. For the experiment, closerange photogrammetric imagery of an in-house plant is used. The results show that the performance of self-supervised MC-CNN is superior to the Census algorithm and comparable to MC-CNN trained by a Light Detection and Ranging point cloud. Another experiment is performed …

AB - Stereo matching can provide complete and dense three-dimensional reconstruction to study plant growth. Recently, high-quality stereo matching results were achieved combining Semi-Global Matching (SGM) with deep learning. However, due to a lack of suitable training data, this technique is not readily applicable for plant reconstruction. We propose a self-supervised Matching Cost with a Convolutional Neural Network (MC-CNN) scheme to calculate matching cost and test it for plant reconstruction. The MC-CNN network is retrained using the initial matching results obtained from the standard MC-CNN weights. For the experiment, closerange photogrammetric imagery of an in-house plant is used. The results show that the performance of self-supervised MC-CNN is superior to the Census algorithm and comparable to MC-CNN trained by a Light Detection and Ranging point cloud. Another experiment is performed …

U2 - 10.14358/PERS.85.5.389

DO - 10.14358/PERS.85.5.389

M3 - Article

VL - 85

SP - 389

EP - 399

JO - Photogrammetric engineering & remote sensing

JF - Photogrammetric engineering & remote sensing

SN - 0099-1112

IS - 5

ER -

Access to Document