defect classification
Recently Published Documents


TOTAL DOCUMENTS

520
(FIVE YEARS 164)

H-INDEX

29
(FIVE YEARS 7)

Author(s):  
Thomas R. Lechtenberg ◽  
Elif Elcin Gunay ◽  
Chih-Yuan Chu ◽  
Gül E. Kremer ◽  
Paul A. Kremer ◽  
...  

2021 ◽  
Vol 38 (6) ◽  
pp. 1783-1791
Author(s):  
Ali Arshaghi ◽  
Mohsen Ashourin ◽  
Leila Ghabeli

Using machine vision and image processing as a non-destructive and rapid method can play an important role in examining defects of agricultural products, especially potatoes. In this paper, we propose a convolution neural network (CNN) to classify the diseased potato into five classes based on their surface image. We trained and tested the developed CNN using a database of 5000 potato images. We compared the results of potato defect classification based on CNN with the traditional neural network and Support Vector Machine (SVM). The results show that the accuracy of the deep learning method is higher than the Traditional Method. We get 100% and 99% accuracy in some of the classes, respectively.


Author(s):  
Yang Liu ◽  
Yachao Yuan ◽  
Jing Liu

Abstract Automatic defect classification is vital to ensure product quality, especially for steel production. In the real world, the amount of collected samples with labels is limited due to high labor costs, and the gathered dataset is usually imbalanced, making accurate steel defect classification very challenging. In this paper, a novel deep learning model for imbalanced multi-label surface defect classification, named ImDeep, is proposed. It can be deployed easily in steel production lines to identify different defect types on the steel's surface. ImDeep incorporates three key techniques, i.e., Imbalanced Sampler, Fussy-FusionNet, and Transfer Learning. It improves the model's classification performance with multi-label and reduces the model's complexity over small datasets with low latency. The performance of different fusion strategies and three key techniques of ImDeep is verified. Simulation results prove that ImDeep accomplishes better performance than the state-of-the-art over the public dataset with varied sizes. Specifically, ImDeep achieves about 97% accuracy of steel surface defect classification over a small imbalanced dataset with a low latency, which improves about 10% compared with that of the state-of-the-art.


2021 ◽  
Vol 11 (23) ◽  
pp. 11459
Author(s):  
Shiqing Wu ◽  
Shiyu Zhao ◽  
Qianqian Zhang ◽  
Long Chen ◽  
Chenrui Wu

The classification of steel surface defects plays a very important role in analyzing their causes to improve manufacturing process and eliminate defects. However, defective samples are very scarce in actual production, so using very few samples to construct a good classifier is a challenge to be addressed. If the layer number of the model with proper depth is increased, the model accuracy will decrease (not caused by overfit), and the training error as well as the test error will be very high. This is called the degradation problem. In this paper, we propose to use feature extraction + feature transformation + nearest neighbors to classify steel surface defects. In order to solve the degradation problem caused by network deepening, the three feature extraction networks of Residual Net, Mobile Net and Dense Net are designed and analyzed. Experiment results show that in the case of a small sample number, Dense block can better solve the degradation problem caused by network deepening than Residual block. Moreover, if Dense Net is used as the feature extraction network, and the nearest neighbor classification algorithm based on Euclidean metric is used in the new feature space, the defect classification accuracy can reach 92.33% when only five labeled images of each category are used as the training set. This paper is of some guiding significance for surface defect classification when the sample number is small.


Author(s):  
Dinh-Thuan Dang ◽  
Jing-Wein Wang ◽  
Jiann-Shu Lee ◽  
Chou-Chen Wang

2021 ◽  
Author(s):  
Jingwei Zhang ◽  
Yi Hu ◽  
Hua Ying ◽  
Yuanqing Mao ◽  
Zhenan Zhu ◽  
...  

Abstract Background: Accurately assessing acetabular defects and designing precise and feasible surgical plans are important before hip revision arthroplasty. With the development of three-dimensional printing, rapid prototyping is a novel technique used to print isometric physical object models. We aimed to propose a three-dimensional acetabular bone defect classification system aided with rapid prototyping and evaluated its reliability and validity.Methods: We reviewed 104 consecutive patients who underwent hip revision arthroplasty in our department between January 2014 and December 2019. Forty five of them had rapid prototyping and were included for reliability and validity test. Three doctors retrospectively evaluated bone defects of these 45 patients with this classification and made surgical plans, and repeated it after 2 weeks. The intra- and inter-observer reliability and the validity to surgical records were assessed using intraclass correlation coefficient or Kappa correlation coefficient.Results: The reliability and validity for classification results were high. The mean initial intraclass correlation coefficient for inter-observer reliability was 0.947, which increased to 0.972 when texted second time. As for inter-observer reliability, it ranged from 0.958 to 0.980. The validity showed high Kappa correlation coefficient of 0.951 to 0.967. When considering detailed surgical plans, the reliability and validity were also high with intraclass correlation coefficient and Kappa correlation coefficient all over 0.9.Conclusions: This three-dimensional acetabulum defect classification was of high reliability and convincing validity. With this classification and objective rapid prototyping models, accurate bone defect assessment and reliable surgical plans were achieved. This classification aided with rapid prototyping could serve as a promising tool for surgeons for preoperative evaluation.


2021 ◽  
Vol 23 (11) ◽  
pp. 867-878
Author(s):  
Ms. Shweta Loonkar ◽  
◽  
Dhirendra S. Mishra ◽  
Surya S. Durbha ◽  
◽  
...  

Quality control unit of fabric industry looks for the effective defect detection methodology. The research is required to be done in this area to develop such solution. Various models based on combination of suitable feature extraction, selection and classification approaches need to be experimented out for the same. This paper attempts to experiment and provide such models mainly based on generic wrapper based selection approaches. Widely used broader range of Haralick features are prominently used for detection and classification of defects in this research. It also attempts to identify the suitability of these features based on segmented images provided as an input. The research has been carried on TILDA Dataset consisting of 800 Silk Fabric Images with eight different defects present on it and each carrying 100 images per defect. Models generated using generic wrapper based approach has also been compared with the Gabor Transforms. Then identification of suitable Haralick Features for particular type of defects has been carried out. In this 68% classification accuracy has been achieved using generic wrapper method and 40 % accuracy has been achieved using Gabor Transform with respect to fourteen Haralick Features and seven types of defects.


Sign in / Sign up

Export Citation Format

Share Document