Contrastive self-supervised representation learning framework for metal surface defect detection DOI Creative Commons

Mahe Zabin,

Anika Nahian Binte Kabir, Muhammad Khubayeeb Kabir

et al.

Journal Of Big Data, Journal Year: 2023, Volume and Issue: 10(1)

Published: Sept. 26, 2023

Abstract Automated detection of defects on metal surfaces is crucial for ensuring quality control. However, the scarcity labeled datasets emerging target poses a significant obstacle. This study proposes self-supervised representation-learning model that effectively addresses this limitation by leveraging both and unlabeled data. The proposed was developed based contrastive learning framework, supported an augmentation pipeline lightweight convolutional encoder. effectiveness approach representation evaluated using pretraining dataset created from three benchmark datasets. Furthermore, performance validated NEU surface-defect dataset. results revealed method achieved classification accuracy 97.78%, even with fewer trainable parameters than models. Overall, extracted meaningful representations image data can be employed in downstream tasks steel defect to improve control reduce inspection costs.

Language: Английский

Evaluation and optimisation of pre-trained CNN models for asphalt pavement crack detection and classification DOI Creative Commons

Sandra Matarneh,

Faris Elghaish, Farzad Pour Rahimian

et al.

Automation in Construction, Journal Year: 2024, Volume and Issue: 160, P. 105297 - 105297

Published: Jan. 31, 2024

This study explored the performance of ten pre-trained CNN architectures in detecting and classifying asphalt pavement cracks from images. A comparison eight optimisation techniques led to developing an optimised model tailored for crack classification, with DenseNet201 emerging as most effective, closely followed by ShuffleNet ResNet101. Conversely, VGG16 exhibited notably lower accuracy among models evaluated. Through application diverse feature selection optimisers, consistently outperformed others, DarkNet19 Xception. Despite employing different VGG19 demonstrated inferior performance. The research introduced a novel approach utilising GWO optimiser validated against various models. Its robustness was verified testing images contaminated differing levels types noise, yielding promising outcomes. Results underscore method's potential accurately types, implying applicability real-world scenarios.

Language: Английский

Citations

28
Deep learning-based models for environmental management: Recognizing construction, renovation, and demolition waste in-the-wild DOI Creative Commons

Diani Sirimewan,

Milad Bazli, Sudharshan N. Raman

et al.

Journal of Environmental Management, Journal Year: 2024, Volume and Issue: 351, P. 119908 - 119908

Published: Jan. 1, 2024

The construction industry generates a substantial volume of solid waste, often destinated for landfills, causing significant environmental pollution. Waste recycling is decisive in managing waste yet challenging due to labor-intensive sorting processes and the diverse forms waste. Deep learning (DL) models have made remarkable strides automating domestic recognition sorting. However, application DL recognize derived from construction, renovation, demolition (CRD) activities remains limited context-specific studies conducted previous research. This paper aims realistically capture complexity streams CRD context. study encompasses collecting annotating images real-world, uncontrolled environments. It then evaluates performance state-of-the-art automatically recognizing in-the-wild. Several pre-trained networks are utilized perform effectual feature extraction transfer during model training. results demonstrated that models, whether integrated with larger or lightweight backbone can composition in-the-wild which useful automated outcome emphasized applicability across various industrial domains, thereby contributing resource recovery encouraging management efforts.

Language: Английский

Citations

23
Improving Industrial Quality Control: A Transfer Learning Approach to Surface Defect Detection DOI Creative Commons
Ângela Semitela, J. M. Dias Pereira, António Completo

et al.

Sensors, Journal Year: 2025, Volume and Issue: 25(2), P. 527 - 527

Published: Jan. 17, 2025

To automate the quality control of painted surfaces heating devices, an automatic defect detection and classification system was developed by combining deflectometry bright light-based illumination on image acquisition, deep learning models for non-defective (OK) defective (NOK) that fused dual-modal information at decision level, online network dispatching visualization. Three decision-making algorithms were tested implementation: a new model built trained from scratch transfer pre-trained networks (ResNet-50 Inception V3). The results revealed two modes employed widened type defects could be identified with this system, while maintaining its lower computational complexity performing multi-modal fusion level. Furthermore, achieved higher accuracies compared to self-built network, ResNet-50 displaying accuracy. inspection consistently obtained fast accurate surface classifications because it imposed OK images both modes. then successfully sent server forwarded graphical user interface showed considerable robustness, demonstrating potential as efficient tool industrial control.

Language: Английский

Citations

4
Computing the characteristics of defects in wooden structures using image processing and CNN DOI
Rana Ehtisham, Waqas Qayyum, Charles V. Camp

et al.

Automation in Construction, Journal Year: 2023, Volume and Issue: 158, P. 105211 - 105211

Published: Nov. 28, 2023

Language: Английский

Citations

30
Automated crack detection and mapping of bridge decks using deep learning and drones DOI
Da Hu, Tien Yee,

Dale Goff

et al.

Journal of Civil Structural Health Monitoring, Journal Year: 2024, Volume and Issue: 14(3), P. 729 - 743

Published: Jan. 8, 2024

Language: Английский

Citations

12
Advancing Crack Detection Using Deep Learning Solutions for Automated Inspection of Metallic Surfaces DOI Creative Commons

Snehal Rathi,

Omkar Mirajkar,

Shubhangi Shukla

et al.

Indian Journal of Information Sources and Services, Journal Year: 2024, Volume and Issue: 14(1), P. 93 - 100

Published: March 30, 2024

The deep Convolutional Neural Network (CNN) architecture used in this research study provides a proof of concept for crack detection on the metallic surface hex nut. goal is to create an automated receiving inspection process supplement human inspections conducted on-site. Conventional image processing techniques (IPTs) have been extensively mechanical infrastructure fault detection. These focus modification extract typical features, such as fractures materials like steel and concrete. However, obstacles presented by variety real-world variables, changes lighting shadows, make it difficult use IPTs. Our suggested vision-based method employs learning CNN overcome these difficulties, eliminating need explicitly compare features. CNNs are more resilient shifting situations than IPTs since they naturally trained identify characteristics images. Following training dataset 1081 images with dimensions 256 x pixels, VGG16 achieved impressive accuracy around 94.17%. Additional architectures, including ResNet, MobileNet, AlexNet, LeNet-5, employed assess accuracies order select most appropriate model. To evaluate robustness flexibility various situations, we tests 206 from alternative structure that was not part dataset. depicted range circumstances, intense light patches tiny fissures. outcomes showed our proposed outperforms current approaches, highlighting its usefulness practical involving identification defects.

Language: Английский

Citations

9
Efficient hybrid ensembles of CNNs and transfer learning models for bridge deck image-based crack detection DOI
Ali Mayya, Nizar Faisal Alkayem, Lei Shen

et al.

Structures, Journal Year: 2024, Volume and Issue: 64, P. 106538 - 106538

Published: May 14, 2024

Language: Английский

Citations

9
Semantic segmentation and deep CNN learning vision-based crack recognition system for concrete surfaces: development and implementation DOI
Yassir M. Abbas,

Hussam Alghamdi

Signal Image and Video Processing, Journal Year: 2025, Volume and Issue: 19(4)

Published: Feb. 24, 2025

Language: Английский

Citations

1
Concrete forensic analysis using deep learning-based coarse aggregate segmentation DOI

Mati Ullah,

Junaid Mir, Syed Sameed Husain

et al.

Automation in Construction, Journal Year: 2024, Volume and Issue: 162, P. 105372 - 105372

Published: March 15, 2024

Language: Английский

Citations

8
A Deep Learning Approach for Surface Crack Classification and Segmentation in Unmanned Aerial Vehicle Assisted Infrastructure Inspections DOI Creative Commons

Shamendra Egodawela,

Amirali K. Gostar,

H. A. D. Samith Buddika

et al.

Sensors, Journal Year: 2024, Volume and Issue: 24(6), P. 1936 - 1936

Published: March 18, 2024

Surface crack detection is an integral part of infrastructure health surveys. This work presents a transformative shift towards rapid and reliable data collection capabilities, dramatically reducing the time spent on inspecting infrastructures. Two unmanned aerial vehicles (UAVs) were deployed, enabling capturing images simultaneously for efficient coverage structure. The suggested drone hardware especially suitable inspection with confined spaces that UAVs broader footprint are incapable accessing due to lack safe access or positioning data. collected image analyzed using binary classification convolutional neural network (CNN), effectively filtering out containing cracks. A comparison state-of-the-art CNN architectures against novel layout “CrackClassCNN” was investigated obtain optimal classification. Segment Anything Model (SAM) employed segment defect areas, its performance benchmarked manually annotated images. achieved accuracy rate 95.02%, SAM segmentation process yielded mean Intersection over Union (IoU) score 0.778 F1 0.735. It concluded selected UAV platform, communication network, processing techniques highly effective in surface detection.

Language: Английский

Citations

6