A Machine Learning Method for Detection of Surface Defects on Ceramic Tiles Using Convolutional Neural Networks

Abstract

We propose a simple but effective convolutional neural network to learn the similarities between closely related raw pixel images for feature representation extraction and classification through the initialization of convolutional kernels from learned filter kernels of the network. The binary-class classification of sigmoid and discriminative feature vectors are simultaneously learned together contrasting the handcrafted traditional method of feature extractions, which split feature-extraction and classification tasks into two different processes during training. Relying on the high-quality feature representation learned by the network, the classification tasks can be efficiently conducted. We evaluated the classification performance of our proposed method using a collection of tile surface images consisting of cracked surfaces and no-cracked surfaces. We tried to classify the tiny-cracked surfaces from non-crack normal tile demarcations, which could be useful for automated visual inspections that are labor intensive, risky in high altitudes, and time consuming with manual inspection methods. We performed a series of comparisons on the results obtained by varying the optimization, activation functions, and deployment of different data augmentation methods in our network architecture. By doing this, the effectiveness of the presented model for smooth surface defect classification was explored and determined. Through extensive experimentation, we obtained a promising validation accuracy and minimal loss.

Surface Defect Detection in Sanitary Ceramics Based on Lightweight Object Detection Network

“…Li et al [9] proposed a two-stage defect detection method, which uses shared features in the defect discovery and defect classification stages to achieve efficient defect detection. Stephen et al [10] proposed a simple CNN model for the detection of ceramic tiles surface defects. In the detection experiment, the method achieved a classification accuracy of 99.43%, but for high-resolution input images, resulting in a long detection time, 16 s.…”

A Hard Voting Policy-Driven Deep Learning Architectural Ensemble Strategy for Industrial Products Defect Recognition and Classification

“…One such intelligent method is the artificial deep learning method that has recently drawn the significant interest of researchers. Deep learning [ 5 ] has been impactful in different research fields involving object detection [ 6 ], object segmentation [ 7 ], image classification [ 8 , 9 ], and speech recognition [ 10 ]. Deep learning provides more efficient and autonomous feature extraction than the manual feature extraction method and thus serves as a substitute in recent related scientific works.…”

Predicting mechanical properties of ultrahigh temperature ceramics using machine learning

“…Several studies 27–29 have developed computer vision‐based ML models in conjunction with nondestructive testing to recognize the surface defects of ceramics. To cite an instance, convolutional neural networks have shown reliable performance on identifying surface cracks and quantifying crack types from ceramic raw images 27 . Some studies have investigated applications of ML on UHTCs.…”

“…Kaufmann et al 26 have leveraged the capability of the random forest (RF) model to discover new chemical compositions for high-entropy ceramics. Several studies [27][28][29] have developed computer vision-based ML models in conjunction with nondestructive testing to recognize the surface defects of ceramics. To cite an instance, convolutional neural networks have shown reliable performance on identifying surface cracks and quantifying crack types from ceramic raw images.…”

“…To cite an instance, convolutional neural networks have shown reliable performance on identifying surface cracks and quantifying crack types from ceramic raw images. 27 Some studies have investigated applications of ML on UHTCs. Qu et al 30 have applied the artificial neural network to predict the melting temperature of UHTCs in relation to chemical compositions.…”