Search research articles

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by

Same author

Molecular Dynamics Analysis of Silica/PMMA Interface Shear Behavior.

Polymers·2022

Same author

Imaging local soil kinematics during the first days of maize root growth in sand.

Scientific reports·2021

Same author

Substrate and cell fusion influence on slime mold network dynamics.

Scientific reports·2021

Same author

Modeling root system growth around obstacles.

Scientific reports·2020

Same author

Transportation networks inspired by leaf venation algorithms.

Bioinspiration & biomimetics·2020

Same author

Substrate composition directs slime molds behavior.

Scientific reports·2019

ABOUT JoVE

Overview Leadership Blog JoVE Help Center

AUTHORS

Publishing Process Editorial Board Scope & Policies Peer Review FAQ Submit

LIBRARIANS

Testimonials Subscriptions Access Resources Library Advisory Board FAQ

RESEARCH

JoVE Journal Methods Collections JoVE Encyclopedia of Experiments Archive

EDUCATION

JoVE Core JoVE Business JoVE Science Education JoVE Lab Manual Faculty Resource Center Faculty Site

Terms & Conditions of Use

Search research articles

Related Experiment Video

Updated: Jun 28, 2025

Determining 3D Flow Fields via Multi-camera Light Field Imaging

Determining 3D Flow Fields via Multi-camera Light Field Imaging

Published on: March 6, 2013

Three-dimensional biphase fabric estimation from 2D images by deep learning.

Daniel Chou¹, Matias Etcheverry², Chloé Arson³

¹Georgia Institute of Technology, School of Civil and Environmental Engineering, Atlanta, GA, 30332-0355, USA.

Scientific Reports

|April 18, 2024

Summary

This summary is machine-generated.

A custom VGG16 model accurately predicts 3D fabric descriptors from 2D images, outperforming a VGG19 model. This approach offers an efficient method for microstructure analysis.

Keywords:

3D fabric descriptor Convolutional neural network Loss function Microstructure analysis Stacked 2D images

More Related Videos

Label-Free Identification of Lymphocyte Subtypes Using Three-Dimensional Quantitative Phase Imaging and Machine Learning

Label-Free Identification of Lymphocyte Subtypes Using Three-Dimensional Quantitative Phase Imaging and Machine Learning

Published on: November 19, 2018

Three-Dimensional Phase Resolved Functional Lung Magnetic Resonance Imaging

Three-Dimensional Phase Resolved Functional Lung Magnetic Resonance Imaging

Published on: June 21, 2024

Related Experiment Videos

Last Updated: Jun 28, 2025

Determining 3D Flow Fields via Multi-camera Light Field Imaging

Determining 3D Flow Fields via Multi-camera Light Field Imaging

Published on: March 6, 2013

Label-Free Identification of Lymphocyte Subtypes Using Three-Dimensional Quantitative Phase Imaging and Machine Learning

Label-Free Identification of Lymphocyte Subtypes Using Three-Dimensional Quantitative Phase Imaging and Machine Learning

Published on: November 19, 2018

Three-Dimensional Phase Resolved Functional Lung Magnetic Resonance Imaging

Three-Dimensional Phase Resolved Functional Lung Magnetic Resonance Imaging

Published on: June 21, 2024

Area of Science:

Materials Science
Computational Modeling
Image Analysis

Background:

Predicting 3D fabric descriptors from 2D images is crucial for material characterization.
Deep learning models offer potential for automating this complex analysis.

Purpose of the Study:

To benchmark a pruned VGG19 model with Axial Coronal Sagittal (ACS) convolutions against a custom VGG16 model for predicting 3D fabric descriptors from 2D images.
To evaluate the performance and computational cost of these models with varying numbers of input images.

Main Methods:

Developed and trained a custom ACS-VGG19 and a custom VGG16 model using numerically generated 3D biphase microstructures.
Extracted 2D image slices from 3D microstructures at regular intervals for model training and testing.
Calculated fabric descriptors from 3D microstructures to serve as ground truth.

Main Results:

The custom VGG16 model achieved a Mean Absolute Percentage Error (MAPE) of 2% or less for several key descriptors, including aggregate size, distance to nearest neighbor, aspect ratios, and solidity.
The ACS-VGG19 model showed a MAPE of 2-5% for aggregate size, aspect ratios, and solidity but could not estimate orientations.
Increasing input images improved VGG16 performance up to a cost-ineffective point; VGG16 outperformed VGG19 due to fewer parameters and faster convergence.

Conclusions:

The custom VGG16 model demonstrates superior performance and efficiency in predicting 3D fabric descriptors from 2D images compared to the ACS-VGG19 model.
Both models predict means more accurately than standard deviations, and aggregate volume fraction less accurately than higher-order descriptors.
The VGG16 model shows promise for predicting orientation descriptors, suggesting its utility across various input image orientations.