Search research articles

Search research articles

Related Experiment Video

Updated: Jun 26, 2025

Surrogate Model Development for Digital Experiments in Welding

Surrogate Model Development for Digital Experiments in Welding

Published on: March 28, 2025

Deep convolutional neural network for weld defect classification in radiographic images.

Dayana Palma-Ramírez¹, Bárbara D Ross-Veitía², Pablo Font-Ariosa³

¹Postgraduate Program Doctorate in Applied Computer Engineering School of Computer Engineering. University of Valparaiso. Valparaiso, Chile.

|May 10, 2024

Summary

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

Multi-criteria evaluation and multi-method analysis for appropriately selecting renewable energy sources in Colombia.

MethodsX·2025

Same author

Machine learning regression algorithms to predict emissions from steam boilers.

Heliyon·2024

Same author

Modeling and control of the hydraulic actuator in a ladle furnace.

Heliyon·2022

Same author

New Arc Stability Index for Industrial AC Three-Phase Electric Arc Furnaces Based on Acoustic Signals.

Sensors (Basel, Switzerland)·2020

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

This summary is machine-generated.

A new deep learning model accurately detects weld defects like cracks and pores in radiographic images. This advanced CNN technology enhances construction safety and quality control, even with low-resolution images.

Area of Science:

Engineering
Computer Science
Materials Science

Background:

Weld quality is paramount for structural integrity and safety in construction.
Early detection of weld defects is essential for preventing structural failures.
Machine vision and deep learning offer advanced solutions for automated weld inspection.

Purpose of the Study:

To develop and evaluate a novel Convolutional Neural Network (CNN) model for classifying weld defects in radiographic images.
To improve the accuracy and efficiency of weld quality control processes.
To assess the model's performance across diverse datasets, including low-quality images.

Main Methods:

Implementation of a CNN model leveraging the ResNet50 architecture.
Training and validation using stratified cross-validation, data augmentation, and regularization techniques.

Keywords:

CNNs Classification Radiographic testing Transfer learning Weld defects

More Related Videos

Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications

Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications

Published on: December 15, 2023

A Swin Transformer-Based Model for Thyroid Nodule Detection in Ultrasound Images

A Swin Transformer-Based Model for Thyroid Nodule Detection in Ultrasound Images

Published on: April 21, 2023

Related Experiment Videos

Last Updated: Jun 26, 2025

Surrogate Model Development for Digital Experiments in Welding

Surrogate Model Development for Digital Experiments in Welding

Published on: March 28, 2025

Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications

Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications

Published on: December 15, 2023

A Swin Transformer-Based Model for Thyroid Nodule Detection in Ultrasound Images

A Swin Transformer-Based Model for Thyroid Nodule Detection in Ultrasound Images

Published on: April 21, 2023

Testing the model on three distinct datasets: RIAWELC, GDXray, and a private low-quality image dataset.

Main Results:

The CNN model achieved high classification accuracies: 98.75% on RIAWELC, 90.255% on GDXray, and 75.83% on the low-quality dataset.
Demonstrated robust performance across varied image qualities and dataset characteristics.
Effectively classified four defect types: crack, pore, non-penetration, and no defect.

Conclusions:

The proposed ResNet50-based CNN model is a highly accurate and effective tool for weld defect detection.
The model significantly enhances the efficiency and reliability of quality control in the welding industry.
The approach shows promise for real-world applications, including inspections of low-resolution radiographic images.