Optimizing Defect Detection on Glossy and Curved Surfaces Using Deep Learning and Advanced Imaging Systems DOI Creative Commons
Jong-Su Yoon, Chibuzo Nwabufo Okwuosa,

Nnamdi Chukwunweike Aronwora

et al.

Sensors, Journal Year: 2025, Volume and Issue: 25(8), P. 2449 - 2449

Published: April 13, 2025

The industrial application of artificial intelligence (AI) has witnessed outstanding adoption due to its robust efficiency in recent times. Image fault detection and classification have also been implemented industrially for product defect detection, as well maintaining standards optimizing processes using AI. However, there are deep concerns regarding the latency performance AI glossy curved surface products, their nature reflective surfaces, which hinder adequate capturing defective areas traditional cameras. Consequently, this study presents an enhanced method curvy image data collection a Basler vision camera with specialized lighting KEYENCE displacement sensors, used train learning models. Our approach employed generated from normal two conditions eight algorithms: four custom convolutional neural networks (CNNs), variations VGG-16, ResNet-50. objective was develop computationally efficient model by deploying global assessment metrics evaluation criteria. results indicate that variation ResNet-50, ResNet-50224, demonstrated best overall efficiency, achieving accuracy 97.97%, loss 0.1030, average training step time 839 milliseconds. terms computational it outperformed one CNN models, CNN6-240, achieved 95.08%, 0.2753, 94 milliseconds, making CNN6-240 viable option resource-sensitive environments.

Language: Английский

Optimizing Defect Detection on Glossy and Curved Surfaces Using Deep Learning and Advanced Imaging Systems DOI Creative Commons
Jong-Su Yoon, Chibuzo Nwabufo Okwuosa,

Nnamdi Chukwunweike Aronwora

et al.

Sensors, Journal Year: 2025, Volume and Issue: 25(8), P. 2449 - 2449

Published: April 13, 2025

The industrial application of artificial intelligence (AI) has witnessed outstanding adoption due to its robust efficiency in recent times. Image fault detection and classification have also been implemented industrially for product defect detection, as well maintaining standards optimizing processes using AI. However, there are deep concerns regarding the latency performance AI glossy curved surface products, their nature reflective surfaces, which hinder adequate capturing defective areas traditional cameras. Consequently, this study presents an enhanced method curvy image data collection a Basler vision camera with specialized lighting KEYENCE displacement sensors, used train learning models. Our approach employed generated from normal two conditions eight algorithms: four custom convolutional neural networks (CNNs), variations VGG-16, ResNet-50. objective was develop computationally efficient model by deploying global assessment metrics evaluation criteria. results indicate that variation ResNet-50, ResNet-50224, demonstrated best overall efficiency, achieving accuracy 97.97%, loss 0.1030, average training step time 839 milliseconds. terms computational it outperformed one CNN models, CNN6-240, achieved 95.08%, 0.2753, 94 milliseconds, making CNN6-240 viable option resource-sensitive environments.

Language: Английский

Citations

0