Industrial Laser Welding Defect Detection and Image Defect Recognition Based on Deep Learning Model Developed

Aiming at the problem of the poor robustness of existing methods to deal with diverse industrial weld image data, we collected a series of asymmetric laser weld images in the largest laser equipment workshop in Asia, and studied these data based on an industrial image processing algorithm and deep learning algorithm. The median filter was used to remove the noises in weld images. The image enhancement technique was adopted to increase the image contrast in different areas. The deep convolutional neural network (CNN) was employed for feature extraction; the activation function and the adaptive pooling approach were improved. Transfer Learning (TL) was introduced for defect detection and image classification on the dataset. Finally, a deep learning-based model was constructed for weld defect detection and image recognition. Specific instance datasets verified the model’s performance. The results demonstrate that this model can accurately identify weld defects and eliminate the complexity of manually extracting features, reaching a recognition accuracy of 98.75%. Hence, the reliability and automation of detection and recognition are improved significantly. The research results can provide a theoretical and practical reference for the defect detection of sheet metal laser welding and the development of the industrial laser manufacturing industry.

Download Full-text

Weld Defect Detection and Image Defect Recognition Using Deep Learning Technology

10.21203/rs.3.rs-149365/v1 ◽

2021 ◽

Author(s):

Yu Cheng ◽

HongGui Deng ◽

YuXin Feng ◽

JunJiang Xiang

Keyword(s):

Deep Learning ◽

Transfer Learning ◽

Defect Detection ◽

Learning Algorithm ◽

Image Features ◽

Economic Losses ◽

Weld Defects ◽

Weld Defect ◽

Defect Recognition ◽

Image Defect

Abstract Welding defects not only bring several economic losses to enterprises and individuals but also threatens peoples lives. We propose a deep learning model, where the data-trained deep learning algorithm is employed to detect the weld defects, and the Convolutional Neural Networks (CNNs) are utilized to recognize the image features. The Transfer Learning (TL) is adopted to reduce the training time via simple adjustments and hyperparameter regulations. The designed deep learning-based model is compared with other classic models to prove its effectiveness in weld defect detection and image recognition further. The results show this model can accurately identify weld defects and eliminates the complexity of manually extracting features, reaching a recognition accuracy of 92.54%. Hence, the reliability and automation of detection and recognition is improved signifificantly. Actual application also verififies the effectiveness of TL in weld defect detection and image defect recognition. Therefore, our research results can provide theoretical and practical references for effificient automatic detection of steel plates, cost reduction, and the high-quality development of iron and steel enterprises.Index Terms - convolutional neural network, deep learning, image detect recognition, transfer learning, weld defect detection

Download Full-text

Image data assessment approach for deep learning-based metal surface defect-detection systems

IEEE Access ◽

10.1109/access.2021.3068256 ◽

2021 ◽

pp. 1-1

Author(s):

Hsien-I Lin ◽

Fauzy Satrio Wibowo

Keyword(s):

Deep Learning ◽

Metal Surface ◽

Defect Detection ◽

Surface Defect ◽

Image Data ◽

Detection Systems ◽

Data Assessment ◽

Assessment Approach ◽

Surface Defect Detection

Download Full-text

Reinforcement based Multi-Model Deep Learning Algorithm for Classification

International Journal of Innovative Technology and Exploring Engineering - Special Issue ◽

10.35940/ijitee.b6465.129219 ◽

2020 ◽

Vol 9 (2) ◽

pp. 1704-1710

Keyword(s):

Deep Learning ◽

Maximum Entropy ◽

Learning Algorithm ◽

Dimensional Space ◽

Image Data ◽

Space Policy ◽

Model Classification ◽

Deep Learning Algorithm ◽

Robust Model ◽

Neural Network Algorithm

Nowadays researchers are focused on processing the multi-media data for classifying the queries of end users by using search engines. The hybrid combination of a powerful classifier and deep feature extractor are used to develop a robust model, which is performed in a high dimensional space. In this research, a three different types of algorithms are combined to attain a stochastic belief space policy, where these algorithms include generative adversary modelling, maximum entropy Reinforcement Learning (RL) and belief space planning which leads to develop a multi-model classification algorithm. In the simulation framework, different adversarial behaviours are used to minimize the agent's action predictability, which has resulted the proposed method to attain robustness, while comparing with unmodelled adversarial strategies. The proposed reinforcement based Deep Learning (DL) algorithm can be used as multi-model classification purpose. The single neural network algorithm can perform the classification on text data and image data. The RL learns the appropriate belief space policy from the feature extracted information of the text and image data, the belief space policy is generated based on the maximum entropy computation

Download Full-text

Defect Detection in Striped Images Using a One-Dimensional Median Filter

Applied Sciences ◽

10.3390/app10031012 ◽

2020 ◽

Vol 10 (3) ◽

pp. 1012

Author(s):

Wei-Chen Lee ◽

Pei-Ling Tai

Keyword(s):

Image Processing ◽

Defect Detection ◽

Median Filter ◽

Detection Methods ◽

Manufacturing Processes ◽

Image Binarization ◽

Image Processing Algorithm ◽

One Dimensional ◽

Modern Manufacturing ◽

The Difference

Defect detection is a key element of quality assurance in many modern manufacturing processes. Defect detection methods, however, often involve a great deal of time and manual work. Image processing has become widely used as a means of reducing the required detection time and effort in manufacturing. To this end, this study proposes an image-processing algorithm for detecting defects in images with striped backgrounds—defect types include scratches and stains. In order to detect defects, the proposed method first pre-processes images and rotates them to align the stripes horizontally. Then, the images are divided into two parts: blocks and intervals. For the blocks, a one-dimensional median filter is used to generate defect-free images, and the difference between the original images and the defect-free images is calculated to find defects. For the intervals, defects are identified using image binarization. Finally, the method superposes the results found in the blocks and intervals to obtain final images with all defects marked. This study evaluated the performance of the proposed algorithm using 65 synthesized images and 20 actual images. The method achieved an accuracy of 97.2% based on the correctness of the defect locations. The defects that could not be identified were those whose greyscales were very close to those of the background.

Download Full-text

Dental Caries Classification System Using Deep Learning Based Convolutional Neural Network

Journal of Computational and Theoretical Nanoscience ◽

10.1166/jctn.2020.9295 ◽

2020 ◽

Vol 17 (9) ◽

pp. 4660-4665

Author(s):

L. Megalan Leo ◽

T. Kalpalatha Reddy

Keyword(s):

Neural Network ◽

Deep Learning ◽

Dental Caries ◽

Convolutional Neural Network ◽

Median Filter ◽

Activation Function ◽

Radiographic Examination ◽

Testing Dataset ◽

The People ◽

Deep Learning Model

In the modern times, Dental caries is one of the most prevalent diseases of the teeth in the whole world. Almost 90% of the people get affected by cavity. Dental caries is the cavity which occurs due to the remnant food and bacteria. Dental Caries are curable and preventable diseases when it is identified at earlier stage. Dentist uses the radiographic examination in addition with visual tactile inspection to identify the caries. Dentist finds difficult to identify the occlusal, pit and fissure caries. It may lead to sever problem if the cavity left untreated and not identified at the earliest stage. Machine learning can be applied to solve this issue by applying the labelled dataset given by the experienced dentist. In this paper, convolutional based deep learning method is applied to identify the cavity presence in the image. 480 Bite viewing radiography images are collected from the Elsevier standard database. All the input images are resized to 128–128 matrixes. In preprocessing, selective median filter is used to reduce the noise in the image. Pre-processed inputs are given to deep learning model where convolutional neural network with Google Net inception v3 architecture algorithm is implemented. ReLu activation function is used with Google Net to identify the caries that provide the dentists with the precise and optimized results about caries and the area affected. Proposed technique achieves 86.7% accuracy on the testing dataset.

Download Full-text

A Review on Machine and Deep Learning for Semiconductor Defect Classification in Scanning Electron Microscope Images

Applied Sciences ◽

10.3390/app11209508 ◽

2021 ◽

Vol 11 (20) ◽

pp. 9508

Author(s):

Francisco López de la Rosa ◽

Roberto Sánchez-Reolid ◽

José L. Gómez-Sirvent ◽

Rafael Morales ◽

Antonio Fernández-Caballero

Keyword(s):

Deep Learning ◽

Electron Microscope ◽

Scanning Electron Microscope ◽

Defect Detection ◽

Scoping Review ◽

Semiconductor Manufacturing ◽

Manufacturing Industry ◽

Future Research ◽

Wide Range ◽

Scanning Electron

Continued advances in machine learning (ML) and deep learning (DL) present new opportunities for use in a wide range of applications. One prominent application of these technologies is defect detection and classification in the manufacturing industry in order to minimise costs and ensure customer satisfaction. Specifically, this scoping review focuses on inspection operations in the semiconductor manufacturing industry where different ML and DL techniques and configurations have been used for defect detection and classification. Inspection operations have traditionally been carried out by specialised personnel in charge of visually judging the images obtained with a scanning electron microscope (SEM). This scoping review focuses on inspection operations in the semiconductor manufacturing industry where different ML and DL methods have been used to detect and classify defects in SEM images. We also include the performance results of the different techniques and configurations described in the articles found. A thorough comparison of these results will help us to find the best solutions for future research related to the subject.

Download Full-text

GI-SleepNet: A Highly Versatile Image-based Sleep Classification using a Deep Learning Algorithm

10.21203/rs.3.rs-586116/v1 ◽

2021 ◽

Author(s):

Tian Xiang Gao ◽

Jia Yi Li ◽

Yuji Watanabe ◽

Chi Jung Hung ◽

Akihiro Yamanaka ◽

...

Keyword(s):

Deep Learning ◽

Learning Algorithm ◽

Sleep Stage ◽

Image Data ◽

Sleep Research ◽

Generative Adversarial Network ◽

Adversarial Network ◽

Deep Learning Algorithm ◽

Technical Requirements ◽

Sleep Stage Classification

Abstract Sleep-stage classification is essential for sleep research. Various automatic judgment programs including deep learning algorithms using artificial intelligence (AI) have been developed, but with limitations in data format compatibility, human interpretability, cost, and technical requirements. We developed a novel program called GI-SleepNet, generative adversarial network (GAN)-assisted image-based sleep staging for mice that is accurate, versatile, compact, and easy to use. In this program, electroencephalogram and electromyography data are first visualized as images and then classified into three stages (wake, NREM, and REM) by a supervised image learning algorithm. To increase the accuracy, we adopted GAN and artificially generated fake REM sleep data to equalize the number of stages. This resulted in improved accuracy, and as few as one mouse data yielded significant accuracy. Because of its image-based nature, it is easy to apply to data of different formats, of different species of animals, and even outside of sleep research. Image data can be easily understood by humans, thus especially confirmation by experts is easy when there are some anomalies of prediction. Because deep learning of images is one of the leading fields in AI, numerous algorithms are also available.

Download Full-text

Application of Data-Driven Iterative Learning Algorithm in Transmission Line Defect Detection

Scientific Programming ◽

10.1155/2021/9976209 ◽

2021 ◽

Vol 2021 ◽

pp. 1-9

Author(s):

Yuquan Chen ◽

Hongxing Wang ◽

Jie Shen ◽

Xingwei Zhang ◽

Xiaowei Gao

Keyword(s):

Deep Learning ◽

Transmission Line ◽

Target Detection ◽

Defect Detection ◽

Learning Algorithm ◽

Small Sample ◽

Iterative Learning ◽

Data Driven ◽

Learning Technology ◽

Line Defect

Deep learning technology has received extensive consideration in recent years, and its application value in target detection is also increasing day by day. In order to accelerate the practical process of deep learning technology in electric transmission line defect detection, the current work used the improved Faster R-CNN algorithm to achieve data-driven iterative training and defect detection functions for typical transmission line defect targets. Based on Faster R-CNN, we proposed an improved network that combines deformable convolution and feature pyramid modules and combined it with a data-driven iterative learning algorithm; it achieves extremely automated and intelligent transmission line defect target detection, forming an intelligent closed-loop image processing. The experimental results show that the increase of the recognition of improved Faster R-CNN network combined with data-driven iterative learning algorithm for the pin defect target is 31.7% more than Faster R-CNN. In the future, the proposed method can quickly improve the accuracy of transmission line defect target detection in a small sample and save manpower. It also provides some theoretical guidance for the practical work of transmission line defect target detection.

Download Full-text

Komparasi Kinerja Algoritma C4.5, Gradient Boosting Trees, Random Forests, dan Deep Learning pada Kasus Educational Data Mining

Jurnal Teknologi Informasi dan Ilmu Komputer ◽

10.25126/jtiik.2020742665 ◽

2020 ◽

Vol 7 (4) ◽

pp. 807

Author(s):

Siti Mutrofin ◽

M. Mughniy Machfud ◽

Diema Hernyka Satyareni ◽

Raden Venantius Hari Ginardi ◽

Chastine Fatichah

Keyword(s):

Deep Learning ◽

Random Forests ◽

Loss Function ◽

Learning Algorithm ◽

Activation Function ◽

Superior Performance ◽

Gradient Boosting ◽

Microsoft Excel ◽

C4.5 Algorithm ◽

Academic Year

Penentuan jurusan di SMA Negeri 1 Jogoroto, Jombang, Jawa Timur menggunakan kurikulum 2013, di mana penentuan jurusan siswa tidak hanya melibatkan keinginan siswa, tes peminatan yang dilakukan siswa di SMA pada minggu pertama, tetapi juga dilengkapi dengan nilai siswa semasa di SMP (nilai rapor siswa, nilai Ujian Nasional, serta rekomendasi guru Bimbingan Konseling), rekomendasi orang tua siswa. Selama ini, sekolah menggunakan proses konvensional dalam menentukan jurusan, yaitu menggunakan Microsoft Excel, yang cenderung lama serta rawan akan kekeliruan dalam melakukan penghitungan. Penentuan jurusan ini dilakukan setiap awal ajaran baru pada siswa baru kelas X. Rata-rata setiap tahun, sekolah mengelola siswa sejumlah 290 dengan waktu dan sumber daya manusia yang terbatas. Pada penelitian ini, penggunaan algoritma ID3 tidak cocok karena data bertipe numerik, sedangkan ID3 hanya mampu menggunakan data bertipe nomial maupun polinomial, sehingga diganti algoritma C4.5. Namun, beberapa penelitian mengatakan algoritma C4.5 memiliki kinerja kurang bagus dibandingkan algoritma Gradient Boosting Trees, Random Forests, dan Deep Learning. Untuk itu, dilakukan perbandingan antara keempat metode tersebut untuk melihat keefektifannya dalam menentukan jurusan di SMA. Data yang digunakan pada penelitian ini adalah data penerimaan siswa baru tahun ajaran 2018/2019. Hasil dari penelitian ini menunjukkan jika atribut yang digunakan bertipe polinomial dengan Deep Learning memiliki kinerja paling unggul untuk semua algoritma jika menggunakan fungsi activation ExpRectifier. Sedangkan jika atributnya bertipe numerik, Deep Learning memiliki kinerja paling unggul untuk semua algoritma jika menggunakan fungsi Tanh untuk semua random sampling. Namun, Deep Learning memiliki kinerja paling buruk untuk semua algoritma jika menggunakan loss Function berupa absolut. Abstract In SMAN 1 Jombang, East Java, the process of determining the students’ majors referred to the 2013 curriculum in which not only the students’ own choices and specialization tests conducted in their first week of SMA were considered but also the student’s SMP grades (a report card, UN scores, and counseling teacher’s recommendation) and parents' recommendation. So far, the school had used Microsoft Excel which required a long time to do and was prone to calculation errors in the process of determination. The process was carried out, with limited time and human resources, at the beginning of a new academic year for grade X students, consisting of 290 students on average. In this present research, the use of ID3 algorithm was not suitable because of its numeric data type instead of nominal or polynomial data. Thus, the C4.5 algorithm was applied, instead. However, the performance of C4.5 algorithm was proved lower than the algorithms of Gradient Boosting Trees, Random Forests, and Deep Learning. Hence, a comparison of performance between them was done to see their effectiveness in the process. The data was the list of new students of the academic year 2018/2019. The results showed that if the attributes are polynomial, the Deep Learning algorithm had the best performance when using the ExpRectifier activation function. When they were numeric, Deep Learning has the most superior performance when using the Tanh function. However, Deep Learning has the worst performance when using the loss function in the form of absolute.

Download Full-text