What question did this study set out to answer?

This study aims to improve defect detection in composite materials using a fusion of deep learning and machine learning techniques.

Intelligent defect detection in composite materials: the role of deep learning and machine learning fusion

View Full Paper

Ask AI

Bookmark

Authors

Randomized trial shows improved defect detection in composite materials, suggesting enhanced quality control methods.

This study aims to improve defect detection in composite materials using a fusion of deep learning and machine learning techniques.
Utilized pre-trained convolutional neural networks (DenseNet169, ResNet152) combined with ML algorithms (MLP, KNN) for non-destructive testing.
Employed transfer learning and feature selection methods (Variance Threshold, XGBoost) on the USimgAIST dataset with 7,004 ultrasonic images.
Evaluated model performance in terms of accuracy, sensitivity, and specificity against false positive and negative rates.
The DenseNet169 + MLP model achieved an accuracy of 99.57% and a Cohen’s Kappa of 0.9912.
The model resulted in 2 false negatives and 4 false positives, minimizing undetected defects.
Feature selection maintained high sensitivity (99.71%) and specificity (99.45%), reducing computational complexity.

Mehmet Akif Buzpınar (2026) studied this question.

View Full Paper

Ask AI

Bookmark