Automatic Quality Research Articles

Automatic quality inspection in additive manufacturing using semi-supervised deep learning

Automatic quality inspection in additive manufacturing using semi-supervised deep learning

Wrist Photoplethysmography Signal Quality Assessment for Reliable Heart Rate Estimate and Morphological Analysis.

Photoplethysmographic (PPG) signals are mainly employed for heart rate estimation but are also fascinating candidates in the search for cardiovascular biomarkers. However, their high susceptibility to motion artifacts can lower their morphological quality and, hence, affect the reliability of the extracted information. Low reliability is particularly relevant when signals are recorded in a real-world context, during daily life activities. We aim to develop two classifiers to identify PPG pulses suitable for heart rate estimation (Basic-quality classifier) and morphological analysis (High-quality classifier). We collected wrist PPG data from 31 participants over a 24 h period. We defined four activity ranges based on accelerometer data and randomly selected an equal number of PPG pulses from each range to train and test the classifiers. Independent raters labeled the pulses into three quality levels. Nineteen features, including nine novel features, were extracted from PPG pulses and accelerometer signals. We conducted ten-fold cross-validation on the training set (70%) to optimize hyperparameters of five machine learning algorithms and a neural network, and the remaining 30% was used to test the algorithms. Performances were evaluated using the full features and a reduced set, obtained downstream of feature selection methods. Best performances for both Basic- and High-quality classifiers were achieved using a Support Vector Machine (Acc: 0.96 and 0.97, respectively). Both classifiers outperformed comparable state-of-the-art classifiers. Implementing automatic signal quality assessment methods is essential to improve the reliability of PPG parameters and broaden their applicability in a real-world context.

Rheological characterization of complex fluids through a table-top 3D printer

3D printing is changing the way we conceive, design, and build 3D objects in mechanical, biomedical, aerospace, construction, automotive and maritime industries. In the current work, the nonlinear rheological behaviour of polymer melts is measured through a table-top 3D printer (3D RheoPrinter) that, smartly modified, allows inline investigation of viscosity, extrudate swell and melt fracture. By using a piezoresistive mini-transducer, the innovative system is designed to be applicable to all Fused Deposition Modelling (FDM) 3D printers by a simple and cost-effective modification of a state-of-art nozzle. The measurements of the nonlinear rheological behaviour are compared with traditional, rotational rheology. Two biodegradable polymers, i.e. polylactic acid and polycaprolactone, are investigated as model systems to test the 3D RheoPrinter. The results of the shear viscosity and the first normal stress difference coefficient, as function of shear rate, show a good agreement between the 3D RheoPrinter and rotational rheometer with an error of about 6% for a confidence interval of 96%. Moreover, the 3D RheoPrinter can still be used as 3D printer. In the last part of this work, it is presented a printing test for building 3D structures in which the results show controllable resolution by means of the measured rheological information such as the extrudate swell. The vision of this work is that an inline rheological characterization, possible with the developed 3D RheoPrinter, can enable automatic process optimization and quality assurance to the 3D printing community. The social and scientific impacts of this work are maximized by the cost-efficiency and simplicity of the design that makes it within reach of the general public. The 3D RheoPrinter opens for a rheological experimentation to a broad audience and it offers important insights to bring FDM to the next level of resolution.

Designing and implementing a smart transplanting framework using programmable logic controller and photoelectric sensor

The importance of automatic seedling transplantation in plant-growing companies cannot be overstated. The key step of intelligent transplanting is the effective and precise detection of fresh saplings. To attain good production, dependability, and reproducibility, enterprises use automation extensively. The use of automatic quality control and assurance system results in a procedure that is cost-effective, dependable, and long-lasting. The sapling tray expressing methodology, the unusual disc comparison four-bar duck mouthpiece seeding method, the electric detector for plantlets detection and classification of a seedling plate, the motion detector, stepping motor, and Programmable Logic Controllers (PLC) management system are all used in the creation of an intelligent different planting structure. It could be used to determine if the soil foundation is filled or not using computer logic programming and management, as well as to regulate the automatic seedling plate transportation. The tray is shifted to minimize scooping up vacant soil foundation by skipping the manipulator’s operation path. Jalapeno plants were used to evaluate the system in an indoor environment. The analysis revealed that if the transplantation speed was 90 min per minute, the percentage of successfully seedling harvesting was 88.24% as well as the percentage of leaking was 17.47%, which met the jalapeno transplantation requirements. The annual percentage of leaks was decreased by around 12% relative to the transplantation model without seedlings recognition. The study serves as a valuable resource for the advancement of smart agricultural transplantation techniques and infrastructure.

Unreferenced English articles' translation quality-oriented automatic evaluation technology using sparse autoencoder under the background of deep learning.

Currently, both manual and automatic evaluation technology can evaluate the translation quality of unreferenced English articles, playing a particular role in detecting translation results. Still, their deficiency is the lack of a close or noticeable relationship between evaluation time and evaluation theory. Thereupon, to realize the automatic Translation Quality Assessment (TQA) of unreferenced English articles, this paper proposes an automatic TQA model based on Sparse AutoEncoder (SAE) under the background of Deep Learning (DL). Meanwhile, the DL-based information extraction method employs AutoEncoder (AE) in the bilingual words’ unsupervised learning stage to reconstruct the translation language vector features. Then, it imports the translation information of unreferenced English articles into Bilingual words and optimizes the extraction effect of language vector features. Meantime, the translation language vector feature is introduced into the automatic DL-based TQA. The experimental findings corroborate that when the number of sentences increases, the number of actual translation errors and the evaluation scores of the proposed model increase, but the Bilingual Evaluation Understudy (BLEU) score is not significantly affected. When the number of sentences increases from 1,000 to 6,000, the BLEU increases from 96 to 98, which shows that the proposed model has good performance. Finally, the proposed model can realize the high-precision TQA of unreferenced English articles.

Stochastic Differential Equations for Automatic Quality Control of Retinal Optical Coherence Tomography images.

Optical coherence tomography is widely used to provide high resolution images from retina. During data acquisition, several artifacts may be associated with OCT images which clearly remove information of retinal layers and degrade the quality of images. Manual assessment of the acquired OCT images is hard and time consuming. Therefore, an automatic quality control step is necessary to detect poor images for next decisions of eliminating them and even re-scanning. In this study, a novel automatic quality control methodology is proposed for early assessment of the OCT images quality by employing stochastic differential equations (SDE). In this method α-stable nature of OCT images is represented by applying a fractional Laplacian filter and parameters of the obtained α-stable are fed to an SVM to automatically detect high quality vs poor quality images. The simulation results on a large dataset of normal and abnormal OCT images show that proposed method has outstanding performance in detection of poor vs high quality images. The methodology is applicable to the image quality assessment of other OCT scanning devices as well. Clinical Relevance- Automatic quality control assessment of retinal OCT images provides reliable data for diagnosis of retinal and systematic diseases in clinical applications.

Pomegranate Fruit Disease Detection Based on Machine Learning

This project presents the recent development in automatic vision-based technology. Use of this technology is increasing in agriculture and fruit industry. An automatic fruit quality detection system for sorting and grading of fruits and defected fruit detection discussed here. The main aim of this system is to replace the manual inspection system. This helps in speed up the process improve accuracy and efficiency and reduce time. This system collect image from camera which is placed on conveyor belt. Then image processing is done to get required features of fruits such as color and size. Defected fruit is detected based on image pixels. Sorting is done based on color and size.

Predicting Age Groups using Brain Imaging Quality Data

Brain imaging quality of data is important to confirm their liability of brain imaging studies. Previous literature confirmed that some confounding factors such as movement, age, and gender may impact brain imaging quality. Automatic quality control (QC) applications may not be able to properly calculate their reliability due to confounding factors.There are a few studies on brain imaging quality data and relevant confounding factors such as age or gender.

Daily Precipitation Data for the Mexico City Metropolitan Area from 1930 to 2015

The Metropolitan Zone of Mexico City, as well as the associated basin, includes the territories of Mexico City, some municipalities of the State of Mexico and the state of Hidalgo. In addition, this area is the most densely populated in Mexico. The region is influenced by mid-latitude and tropical weather systems and is vulnerable to extreme hydrometeorological events. In this context, we developed a dataset from the records of 136 geolocated sites that includes daily precipitation data from the CLImate COMputing (CLICOM) project and the Mexico City Water System. The data spans the period from 1930 to 2015 for the rainy months (June–October) from stations with records of 20 or more years. In each recording site, automatic and manual data quality control were performed to verify the consistency of the daily precipitation data. We believe that our highly dense precipitation dataset will be useful for climate, trend and extreme events analysis. Additionally, the data will allow validating simulations of numerical atmospheric models. The dataset is public, and it was previously used in other research to determine areas susceptible to flooding due to heavy rain events and to develop a web mapping application of daily precipitation data.

Automatic ECG classification and label quality in training data

Objective. Within the PhysioNet/Computing in Cardiology Challenge 2021, we focused on the design of a machine learning algorithm to identify cardiac abnormalities from electrocardiogram recordings (ECGs) with a various number of leads and to assess the diagnostic potential of reduced-lead ECGs compared to standard 12-lead ECGs. Approach. In our solution, we developed a model based on a deep convolutional neural network, which is a 1D variant of the popular ResNet50 network. This base model was pre-trained on a large training set with our proposed mapping of original labels to SNOMED codes, using three-valued labels. In the next phase, the model was fine-tuned for the Challenge metric and conditions. Main results. In the Challenge, our proposed approach (team CeZIS) achieved a Challenge test score of 0.52 for all lead configurations, placing us 5th out of 39 in the official ranking. Our improved post-Challenge solution was evaluated as the best for all ranked configurations, i.e. for 12-lead, 3-lead, and 2-lead versions of the full test set with the Challenge test score of 0.62, 0.61, and 0.59, respectively. Significance. In addition to building the model for identifying cardiac anomalies, we provide a more detailed description of the issues associated with label mapping and propose its modification in order to obtain a better starting point for training more powerful classification models. We compare the performance of models for different numbers of leads and identify labels for which two leads are sufficient. Moreover, we evaluate the label quality in individual parts of the Challenge training set.

Spatiotemporal Variation in Air Pollution Characteristics and Influencing Factors in Ulaanbaatar from 2016 to 2019

Ambient air pollution is a global environmental issue that affects human health. Ulaanbaatar (UB), the capital of Mongolia, is one of the most polluted cities in the world, and it is of great importance to study the temporal and spatial changes in air pollution in this city, along with their influencing factors. To understand the characteristics of atmospheric pollutants in UB, the contents of PM10, PM2.5, SO2, NO2, CO, and O3, as well as their influencing factors, were analyzed from data obtained from automatic air quality monitoring stations. These analyses yielded six major findings: (1) From 2016 to 2019, there was a total of 883 pollution days, and PM2.5 and PM10 were the primary pollutants on 553 and 351 of these days, respectively. The air pollution was dominated by PM10 in spring and summer, affected by both PM2.5 and PM10 in autumn, and dominated by PM2.5 in winter. (2) Compared with 2016, the number of days with good air quality in UB in 2019 increased by 45%, and the number of days with unhealthy or worse levels of pollution decreased by 56%, indicating that the air quality improved year by year. (3) From 2016 to 2019, the annual average PM2.5/PM10 ratio dropped from 0.55 to 0.45, and the proportion of PM2.5 in particulate matter decreased year by year. The PM concentration and PM2.5/PM10 ratio were highest in winter and lowest in summer. When comparing the four-season averages, the average PM2.5 concentration decreased by 89% from its highest level, and the PM10 concentration decreased by 67%, indicating stronger seasonal differences in PM2.5 than in PM10. (4) The hourly changes in PM concentration showed a bimodal pattern, exhibiting a decrease during the day and a slight increase in the afternoon due to temperature inversion, so the PM2.5/PM10 ratio increased at night in all four seasons. The PM concentration during the heating season was significantly higher than that in the non-heating season, indicating that coal-fired heating was the main cause of air pollution in UB. (5) Sand dust and soot were the two main types of pollution in UB. (6) Correlation analysis and linear fitting analysis showed that PM2.5 and PM10 caused by coal-firing had an important impact on air quality in UB. Coal combustion and vehicle emissions with SO2, NO2, and CO as factors made large contributions to PM2.5.

Automated image quality appraisal through partial least squares discriminant analysis.

Automatic retinal fundus image quality analysis is one of the most essential preliminary stages in automatic computer-aided retinal disease diagnosis system, which allows good-quality fundus images for accurate disease prediction through localization and segmentation of retinal regions. This paper presents new feature extraction methods using full-reference and no-reference image quality metrics for image quality classification. Basic image features, reference and no-reference features are extracted from the fundus image and applied through different classification techniques to determine the image quality for further diagnosis. In this paper, human-made categorization including good and non-good-quality fundus image classification is constructed by considering major features of retinal fundus images are illumination, clarity, image intensity, contrast and region visibility. The proposed system presented fundus image quality classification by automatic extraction of features from fundus images through image processing techniques and automatic classification of image quality through different classification algorithm. This system was thoroughly investigated on 2674 retinal fundus images from publically available datasets, namely MESSIDOR, Drishti-GS1, DRIVE, HRF, DRIONS-DB, DIARETDB0, DIARETDB1, IDRiD, INSPIRE-AVR, CHASE-DB1, ONHSD, DRIMDB and e-ophtha-EX with better performance results in terms of sensitivity, accuracy, precision and F1 score of 99.36%, 96.79%, 96.29% and 97.79%, respectively. The proposed system results were compared to the existing state-of-the-art approaches and outperform existing methods for image quality assessment representing the efficiency and robustness of our system is most suitable for automatic image analysis during retinal disease diagnosis.

DEVELOPING DESIGN OF AUTOMATIC EGG QUALITY DETECTOR USING ROI AND RGB TEMPLATE METHODS

The eggs of purebred chickens always become the major choice for satisfying people’s needs for nutrients. The way to process them is easy, and the prices are relatively cheap for all people. Since the eggs of purebred chickens are in high demand, a device that can detect their qualities more correctly and accurately is needed, compared to the conventional method. Objective: An automatic egg quality detector was developed through a camera module of pixy cmucam5 by employing ROI and template RGB methods. The implementation of the ROI method could optimize the system performance, thereby accelerating the process of egg quality detection carried out by the preprocessor of the pixy camera. ROI analyzed the egg image only on the desired pixel area, not the whole pixel. As a result, the ROI method yielded a device and system that could detect the egg quality within the average speed of detection time -/+ 2ms per egg out of 4 trial categories. Meanwhile, the template RGB method was able to improve and ease the ability of the pixy camera module in detecting the egg quality. When the RGB index of the egg was < 234, then the egg was of good quality. Contrarily, when the RGB index of the egg was > 234, the egg quality was in a bad category. The contribution of this paper is to develop a low-cost automatic egg sorting and detection system with the advantages of flexibility in device placement, due to the use of camera modules and object detection accuracy, as well as better response time than using color sensors. Accordingly, the template RGB method produced a device and system which could detect egg quality with average accuracy of 90 % and average error of 10% out of 4 trial categories. And, the results of the trials upon the detector system indicated success in detecting the egg quality.

VASCULAR AGING IN THE YOUNG: CAROTID STIFFNESS POPULATION CENTILES AND ASSOCIATION WITH BLOOD PRESSURE IN THE KIGGS COHORT

Objective: Functional characteristics of the vascular system underly changes associated with aging, such as increasing arterial stiffness. Research suggests that these age-associated changes may be accelerated in the presence of cardiovascular risk factors. So far, normative data on arterial stiffness in unselected adolescents and young adults have been scarce. Recently, measurements of carotid stiffness (cS) parameters have been included in a national health examination survey. Design and method: The population-based KiGGS cohort 11-year-follow-up included high-resolution B-mode sonography with semi-automated edge-detection and automatic electrocardiogram-gated real-time quality control. In 4,305 participants aged 14–28 years, distensibility coefficient (DC), stiffness index ß, Young's (YEM) and Peterson's elastic modulus (Ep) were assessed. Centiles were modelled by sex, age and height simultaneously, using generalized additive models for location, scale and shape. Log- binomial regression models were used to investigate associations of blood pressure with cS > = 90th centile (P90) as outcomes. Hypertensive blood pressure was defined according to German guidelines (> = P95 systolic or diastolic for age, height and sex according to KiGGS centiles until age 17 and > = 140/90 mmHg from age 18). Results: Multivariable models show that hypertensive blood pressure in childhood is associated with an increased risk of elevated cS (> = P90 for age, height and sex) eleven years later for all parameters of cS, with the exception of stiffness index ß (no association). Effect estimates for DC, YEM and Ep range from 1.77 [CI 1.28–2.44] for DC to 1.41 [CI 0.99–2.01] for YEM. Cross-sectional associations of blood pressure with cS parameters show even stronger associations (relative risks between 1.3 and 3.3). Conclusions: Using state-of-the-art sonography and centiles computed with novel statistical methods, this study confirms that, at the population level, hypertensive blood pressure in childhood is linked to increased carotid stiffness in young adulthood. The results emphasize that healthy lifestyles and healthy living environments in childhood and adolescence represent an early investment in cardiovascular health.

AIVMAF: Automatic Image Quality Estimation Based on Improved VMAF and YOLOv4

The current most widely used way of image quality estimation relies heavily on the subjective assessment, while majority of past objective estimation methods are not satisfactory on accuracy. To solve them and realize unsupervised image quality estimation with high precision, this paper creates a linear way with “Proportional Partition” controlled by horizontal and vertical rates of extracted pixel to get best representations of the image with patching, balance the uneven distribution of image quality in each source image, and offer dynamic compatibility to images with high resolution. Besides, it estimates the image quality automatically with a model trained by current best artificial intelligence (AI) algorithm for target detection YOLOv4 with 1000 images random selected from ImageNet2013 database. The proposal also uses the spirit of joint indices from the current widely used method named Video Multimethod Assessment Fusion (VMAF). But we replace its Visual Information Fidelity (VIF) with Visual Saliency-induced Index (VSI) and add VSI to our target function because of VIF’s dependence on subjective assessment, and also for VSI’s better performance surpassing most recent IQA estimators as TOP3 best model in recent world. Besides, contrast masking is also included by objective function for the KL-divergence to simulate the human visual perception better. A creative “Batch Learning” way is found to address patches for less calculation and faster speed. All source images are pretreated with colour space transformation and normalization to improve descriptiveness of images and reduce the redundant points, and a threshold is devised to formulate suppression mechanisms. The proposed solution is tested to be a good image quality assessor in many aspects such as correctness, consistency, linearity, monotonicity and speed, and performs well on even HD images.

Yolo V4 딥러닝 지능기술을 이용한 과일 불량 부위 검출

It is very important to first detect and remove defective fruits with scratches or bruised areas in the automatic fruit quality screening system. This paper proposes a method of detecting defective areas in fruits using the latest artificial intelligence technology, the Yolo V4 deep learning model in order to overcome the limitations of the method of detecting fruit’s defective areas using the existing image processing techniques. In this study, a total of 2,400 defective fruits, including 1,000 defective apples and 1,400 defective fruits with scratch or decayed areas, were learned using the Yolo V4 deep learning model and experiments were conducted to detect defective areas. As a result of the performance test, the precision of apples is 0.80, recall is 0.76, IoU is 69.92% and mAP is 65.27%. The precision of pears is 0.86, recall is 0.81, IoU is 70.54% and mAP is 68.75%. The method proposed in this study can dramatically improve the performance of the existing automatic fruit quality screening system by accurately selecting fruits with defective areas in real time rather than using the existing image processing techniques.

Evaluation of the Accuracy of Artificial Intelligence Translation Based on Deep Learning

A machine automatic translation quality evaluation model based on a deep learning algorithm is built with the aim of achieving an accurate evaluation of machine automatic translation quality. Deep learning-based automatic machine translation was made. During the unsupervised and supervised learning stages, language information extraction executes unsupervised learning and reconstructs automated translation samples of bilingual words using a noise lessening automatic coding machine. To improve the effect of language vector feature extraction, the machine optimizes the effect of language vector feature extraction by importing language vector function and machine automatic translation information into bilingual words. The machine automatic translation language vector function is imported into the translation quality evaluation model based on deep learning to realize the automatic machine translation quality evaluation. The experimental results show that the constructed model can accurately evaluate the quality of machine automatic translation, the pattern of translated sentences and the number of sentences do not adversely affect the evaluation performance of the model, and the model exerts a good effect is showing.

Blind image quality assessment of magnetic resonance images with statistics of local intensity extrema

Magnetic resonance (MR) imaging provides a large amount of data that requires a visual inspection before a diagnosis can be made. Since the exclusion of low-quality image sequences is performed manually and image processing methods are evaluated using techniques developed for natural images, automatic and reliable MR image quality assessment (IQA) approaches are desirable. Therefore, in this work, a new no-reference (NR) MR-IQA technique is proposed. The method uses introduced quality-aware features addressing characteristics of MR images. Specifically, in the method, an MR image is scaled, filtered with two gradient operators, and subjected to identification of the local intensity extrema. Then, the entropy and κ curvature are calculated to characterize extrema sequences and used as perceptual features to train a quality model with the Support Vector Regression (SVR) technique. In this paper, an extensive comparative evaluation of the method against recent NR approaches, including deep learning-based models, is conducted on two representative MR-IQA benchmarks. The results reveal the superiority of the introduced approach over competing methods as it obtained better overall Spearman and Pearson correlation coefficients by 5% and 3%, respectively.

Computation of Strip Road Networks Based on Harvester Location Data

The location information of strip roads in thinnings and the numerical variables of strip roads are one aspect of timber harvest quality information. The ideal of automatic quality management for mechanized logging is that the quality of the harvest is calculated and reported based on data collected by forest machines. At present, quality data is collected by means of laborious, manual, field-based, post-harvest measurements. The aim of this study was to develop an automatic method to compute the strip roads of the harvested stands after harvesting, based on the stem-specific location data of the harvester. Subsequently, the strip road variables were computed from the strip road networks. The computed strip road networks were validated with 21 manually recorded field references. The method was also applied to operational stand data, including 544 harvested stands collected from Southern Finland. The results showed that the computation method produces well-located, stand-specific strip road networks from which strip road variables can be accurately determined, covering the whole stand. Thus, the method promotes the automation of quality management and reporting. The computed strip road networks can also support harvester operator work during the harvesting and, later, the automation of harvesting operations.

ORYX‐MRSI: A fully‐automated open‐source software for proton magnetic resonance spectroscopic imaging data analysis

AbstractProton magnetic resonance spectroscopic imaging (1H‐MRSI) provides a noninvasive, spatially resolved evaluation of brain metabolism. However, there are some limitations of 1H‐MRSI preventing its wider use in the clinics, including the spectral quality issues, partial volume effect and chemical shift artifact. Additionally, it is necessary to create metabolite maps for analyzing spectral data along with other MRI modalities. In this study, a MATLAB‐based open‐source data analysis software for three‐dimensional 1H‐MRSI, called Oryx‐MRSI, which includes modules for visualization of raw 1H‐MRSI data and LCModel outputs, chemical shift correction, tissue fraction calculation, metabolite map production, and registration onto standard MNI152 brain atlas while providing automatic spectral quality control, is presented. Oryx‐MRSI implements region of interest analysis at brain parcellations defined on MNI152 brain atlas. All generated metabolite maps are stored in NIfTI format. Oryx‐MRSI is publicly available at https://github.com/Computational-Imaging-LAB/Oryx-MRSI along with six example datasets.