Rapid Defect Detection and Classification in Images Using Convolutional Neural Networks

2021 ◽  
Author(s):  
Peter Warren ◽  
Hessein Ali ◽  
Hossein Ebrahimi ◽  
Ranajay Ghosh
Keyword(s):  
Machine Learning ◽  
Neural Networks ◽  
Convolutional Neural Networks ◽  
Defect Detection ◽  
Visual Inspection ◽  
Surface Defects ◽  
Image Data ◽  
Rapid Identification ◽  
Machine Learning Techniques ◽  
Great Success

Abstract Several image processing methods have been implemented over recent years to assist and partially replace on-site technician visual inspection of both manufactured parts and operational equipments. Convolutional neural networks (CNNs) have seen great success in their ability to both identify and classify anomalies within images, in some cases they do this to a higher degree of accuracy than an expert human. Several parts that are manufactured for various aspects of turbomachinery operation must undergo a visual inspection prior to qualification. Machine learning techniques can streamline these visual inspection processes and increase both efficiency and accuracy of defect detection and classification. The adoption of CNNs to manufactured part inspection can also help to improve manufacturing methods by rapidly retrieving data for overall system improvement. In this work a dataset of images with a variety of surface defects and some without defects will be fed through varying CNN set-ups for the rapid identification and classification of the flaws within the images. This work will examine the techniques used to create CNNs and how they can best be applied to part surface image data, and determine the most accurate and efficient techniques that should be implemented. By combining machine learning with non-destructive evaluation methods component health can be rapidly determined and create a more robust system for manufactured parts and operational equipment evaluation.

Designing of smart chair for monitoring of sitting posture using convolutional neural networks

2019 ◽  
Vol 53 (2) ◽  
pp. 142-155 ◽  
Author(s):  
Wonjoon Kim ◽  
Byungki Jin ◽  
Sanghyun Choo ◽  
Chang S. Nam ◽  
Myung Hwan Yun
Keyword(s):  
Machine Learning ◽  
Neural Networks ◽  
Convolutional Neural Networks ◽  
Image Data ◽  
Research Area ◽  
Machine Learning Algorithms ◽  
Machine Learning Techniques ◽  
Content Type ◽  
Sitting Posture ◽  
Posture Classification

Purpose Sitting in a chair is a typical act of modern people. Prolonged sitting and sitting with improper postures can lead to musculoskeletal disorders. Thus, there is a need for a sitting posture classification monitoring system that can predict a sitting posture. The purpose of this paper is to develop a system for classifying children’s sitting postures for the formation of correct postural habits. Design/methodology/approach For the data analysis, a pressure sensor of film type was installed on the seat of the chair, and image data of the postu.re were collected. A total of 26 children participated in the experiment and collected image data for a total of seven postures. The authors used convolutional neural networks (CNN) algorithm consisting of seven layers. In addition, to compare the accuracy of classification, artificial neural networks (ANN) technique, one of the machine learning techniques, was used. Findings The CNN algorithm was used for the sitting position classification and the average accuracy obtained by tenfold cross validation was 97.5 percent. The authors confirmed that classification accuracy through CNN algorithm is superior to conventional machine learning algorithms such as ANN and DNN. Through this study, we confirmed the applicability of the CNN-based algorithm that can be applied to the smart chair to support the correct posture in children. Originality/value This study successfully performed the posture classification of children using CNN technique, which has not been used in related studies. In addition, by focusing on children, we have expanded the scope of the related research area and expected to contribute to the early postural habits of children.

scholarly journals Automating Visual Inspection with Convolutional Neural Networks

2019 ◽  
Vol 11 (1) ◽  
Author(s):  
Sreerupa Das ◽  
Christopher D Hollander ◽  
Suraiya Suliman
Keyword(s):  
Neural Networks ◽  
Convolutional Neural Networks ◽  
Defect Detection ◽  
Surface Defects ◽  
Training Data ◽  
Subject Matter Experts ◽  
Data Set ◽  
Analysis Task ◽  
Aerial Vehicle ◽  
Initial Results

Convolutional Neural Networks (CNNs) have become the recent tool of choice for many visual detection tasks, including object classification, localization, detection, and segmentation. CNNs are specialized neural networks composed of many layers and specifically designed to analyze grid-like data, e.g. images. One of the key features of a CNN is its ability to automatically detect important features within an image (e.g. edges, patterns, shapes); prior to CNNs, these features had to be manually engineered by subject matter experts. Inspired by the significant achievements and success that CNNs have experienced in the domain of computer vision, we examine a specific convolutional neural network (CNN) architecture, U-Net, suited for the task of visual defect detection. We identify and discuss situations for the use of this architecture in the specific context of external defect detection on aircraft and experimentally discuss its performance across a dataset of common visual defects. One requirement of training Convolution Networks on an image analysis task is the need for a large image (training) data set.  We address this problem by using synthetically generated images from computer models of jets with varying angles and perspectives with and without induced faults in the generated images.  This paper presents the initial results of using CNNs, specifically U-Net, to detect aerial vehicle surface defects of three categories.  We further demonstrate that CNNs trained on synthetic images can then be used to detect faults in real images of jets with visual damages.  The results obtained in this research, indicate that our approach has been quite effective in detecting surface anomalies in our tests.

scholarly journals Steganalysis Using Yedrodj-net net's Convolutional Neural Networks (CNN) Method on Steganography Tools

2020 ◽  
Vol 3 ◽  
pp. 207-211
Author(s):  
Nurmi Hidayasari ◽  
Imam Riadi ◽  
Yudi Prayudi
Keyword(s):  
Machine Learning ◽  
Neural Networks ◽  
Convolutional Neural Networks ◽  
Learning Performance ◽  
Machine Learning Techniques ◽  
Detection Model ◽  
Learning Techniques ◽  
Net Method ◽  
Blind Steganalysis ◽  
Better Than

Steganalysis method is used to detect the presence or absence of steganography files or can be referred to anti-steganography. Steganalysis can be used for positive purposes, which is to know the weaknesses of a steganography method, so that improvements can be made. One category of steganalysis is blind steganalysis, which is a way to detect secret files without knowing what steganography method is used. Blind steganalysis is difficult to implement, but then machine learning techniques emerged that could be used to create a detection model using experimental data, one of which is Convolutional Neural Networks (CNN). A study proposes that the CNN method can detect steganography files using the latest method with a low error probability value compared to other methods, CNN Yedroudj-net. As one of the steganalysis methods with the latest machine learning steganalysis techniques, an experiment is needed to find out whether Yedroudj-net can be a steganalysis for the output of many tools commonly used for steganography applications. Knowing the performance of CNN Yedroudj-net on several steganography tools is very important, to measure the level of ability in terms of steganalysis of some of these tools. Especially so far, machine learning performance is still doubtful in blind steganalysis. Plus some previous research only focused on certain methods to prove the performance of the proposed technique, including Yedroudj-net. This study will use five tools that are Hide In Picture (HIP), OpenStego, SilentEye, Steg and S-Tools, which are not known exactly what steganography methods are used on the tools. Yedroudj-net method will be implemented in the steganography file from the output of the five tools. Then a comparison with the popular steganalysis tool is used, StegSpy. The results show that Yedroudj-net is quite capable of detecting the presence of steganography files, slightly better than StegSpy.

Recognition on Images From Internet Street View Based on Hierarchical Features Learning With CNNs

2019 ◽  
pp. 1411-1424
Author(s):  
Jian-min Liu ◽  
Min-hua Yang
Keyword(s):  
Machine Learning ◽  
Neural Networks ◽  
Unsupervised Learning ◽  
Convolutional Neural Networks ◽  
Error Rates ◽  
Machine Learning Techniques ◽  
Street View ◽  
Learning Techniques ◽  
Similarities And Differences ◽  
Hierarchical Features

This article describes hierarchical features with unsupervised learning on images from internet street view images. This is due to the time spent by trained researchers on feature construction steps with traditional methods. This article focuses on the activation of each layer of with convolutional neural networks (CNNs) on Internet street view images detection and compared similarities and differences among them on each layer. The experiment results achieved error rates of 21% on recognition which work went relatively well than the traditional machine learning techniques, such as Parallel SVM.

scholarly journals An efficient classification of flower images with convolutional neural networks

2017 ◽  
Vol 7 (1.1) ◽  
pp. 384 ◽  
Author(s):  
M V.D. Prasad ◽  
B JwalaLakshmamma ◽  
A Hari Chandana ◽  
K Komali ◽  
M V.N. Manoja ◽  
...  
Keyword(s):  
Artificial Intelligence ◽  
Machine Learning ◽  
Neural Networks ◽  
Convolutional Neural Networks ◽  
Recognition Rate ◽  
Image Data ◽  
Image Database

Machine learning is penetrating most of the classification and recognition tasks performed by a computer. This paper proposes the classification of flower images using a powerful artificial intelligence tool, convolutional neural networks (CNN). A flower image database with 9500 images is considered for the experimentation. The entire database is sub categorized into 4. The CNN training is initiated in five batches and the testing is carried out on all the for datasets. Different CNN architectures were designed and tested with our flower image data to obtain better accuracy in recognition. Various pooling schemes were implemented to improve the classification rates. We achieved 97.78% recognition rate compared to other classifier models reported on the same dataset.

A New Student Modeling Technique With Convolutional Neural Networks: LearnerPrints

2020 ◽  
pp. 073563312096921
Author(s):  
Şeyhmus Aydoğdu
Keyword(s):  
Machine Learning ◽  
Neural Networks ◽  
Convolutional Neural Networks ◽  
Management System ◽  
Learning Management System ◽  
Modeling Technique ◽  
Machine Learning Techniques ◽  
Student Modeling ◽  
Student Model ◽  
Learning Techniques

Student modeling is one of the most important processes in adaptive systems. Although learning is individual, a model can be created based on patterns in student behavior. Since a student model can be created for more than one student, the use of machine learning techniques in student modeling is increasing. Artificial neural networks (ANNs), which form one group of machine learning techniques, are among the methods most frequently used in learning environments. Convolutional neural networks (CNNs), which are specific types of these networks, are used effectively for complex problems such as image processing, computer vision and speech recognition. In this study, a student model was created using a CNN due to the complexity of the learning process, and the performance of the model was examined. The student modeling technique used was named LearnerPrints. The navigation data of the students in a learning management system were used to construct the model. Training and test data were used to analyze the performance of the model. The classification results showed that CNNs can be used effectively for student modeling. The modeling was based on the students’ achievement and used the students’ data from the learning management system. The study found that the LearnerPrints technique classified students with an accuracy of over 80%.

scholarly journals Image Completion using Spiking Neural Networks

2019 ◽  
Vol 9 (1) ◽  
pp. 4449-4452
Keyword(s):  
Machine Learning ◽  
Neural Networks ◽  
Convolutional Neural Networks ◽  
Spiking Neural Networks ◽  
Machine Learning Techniques ◽  
Free Form ◽  
Third Generation ◽  
Image Completion ◽  
Learning Techniques ◽  
Original Picture

In this paper, we are showing how spiking neural networks are applied in image repainting, and its results are outstanding compared with other machine learning techniques. Spiking Neural Networks uses the shape of patterns and shifting distortion on images and positions to retrieve the original picture. Thus, Spiking Neural Networks is one of the advanced generations and third generation of machine learning techniques, and is an extension to the concept of Neural Networks and Convolutional Neural Networks. Spiking Neural Networks (SNN) is biologically plausible, computationally more powerful, and is considerably faster. The proposed algorithm is tested on different sized digital images over which free form masks are applied. The performance of the algorithm is examined to find the PSNR, QF and SSIM. The model has an effective and fast to complete the image by filling the gaps (holes).

scholarly journals Jet Features: Hardware-Friendly, Learned Convolutional Kernels for High-Speed Image Classification

Electronics ◽  
2019 ◽  
Vol 8 (5) ◽  
pp. 588 ◽  
Author(s):  
Taylor Simons ◽  
Dah-Jye Lee
Keyword(s):  
Machine Learning ◽  
Neural Networks ◽  
Image Classification ◽  
Convolutional Neural Networks ◽  
High Speed ◽  
Visual Inspection ◽  
Learning Algorithms ◽  
Machine Learning Algorithms ◽  
Inspection Tasks ◽  
Eco Features

This paper explores a set of learned convolutional kernels which we call Jet Features. Jet Features are efficient to compute in software, easy to implement in hardware and perform well on visual inspection tasks. Because Jet Features can be learned, they can be used in machine learning algorithms. Using Jet Features, we make significant improvements on our previous work, the Evolution Constructed Features (ECO Features) algorithm. Not only do we gain a 3.7× speedup in software without loosing any accuracy on the CIFAR-10 and MNIST datasets, but Jet Features also allow us to implement the algorithm in an FPGA using only a fraction of its resources. We hope to apply the benefits of Jet Features to Convolutional Neural Networks in the future.

scholarly journals Identifying Ear Abnormality from 2D Photographs Using Convolutional Neural Networks

2019 ◽  
Vol 9 (1) ◽  
Author(s):  
Rami R. Hallac ◽  
Jeon Lee ◽  
Mark Pressler ◽  
James R. Seaward ◽  
Alex A. Kane
Keyword(s):  
Machine Learning ◽  
Neural Networks ◽  
Deep Learning ◽  
Convolutional Neural Networks ◽  
Machine Learning Techniques ◽  
Test Accuracy ◽  
Linear Measurements ◽  
Softmax Classifier ◽  
Ear Deformity ◽  
Learning Techniques

AbstractQuantifying ear deformity using linear measurements and mathematical modeling is difficult due to the ear’s complex shape. Machine learning techniques, such as convolutional neural networks (CNNs), are well-suited for this role. CNNs are deep learning methods capable of finding complex patterns from medical images, automatically building solution models capable of machine diagnosis. In this study, we applied CNN to automatically identify ear deformity from 2D photographs. Institutional review board (IRB) approval was obtained for this retrospective study to train and test the CNNs. Photographs of patients with and without ear deformity were obtained as standard of care in our photography studio. Profile photographs were obtained for one or both ears. A total of 671 profile pictures were used in this study including: 457 photographs of patients with ear deformity and 214 photographs of patients with normal ears. Photographs were cropped to the ear boundary and randomly divided into training (60%), validation (20%), and testing (20%) datasets. We modified the softmax classifier in the last layer in GoogLeNet, a deep CNN, to generate an ear deformity detection model in Matlab. All images were deemed of high quality and usable for training and testing. It took about 2 hours to train the system and the training accuracy reached almost 100%. The test accuracy was about 94.1%. We demonstrate that deep learning has a great potential in identifying ear deformity. These machine learning techniques hold the promise in being used in the future to evaluate treatment outcomes.

Sign in / Sign up

Export Citation Format

Share Document