Unsupervised Machine Learning Via Transfer Learning and k-Means Clustering to Classify Materials Image Data

Cohn, Ryan; Holm, Elizabeth A.

doi:10.1007/s40192-021-00205-8

Cited by 50 publications

(27 citation statements)

References 34 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…ResNet101 [ 59 ] and DenseNet201 [ 58 ], with 101 and 201 layers, respectively, were used for feature generation [ 58 , 59 ]. Initially, these models were trained using the ImageNet dataset [ 34 ], and, by now, these models have been used extensively for transfer learning applications [ 63 , 64 ]. As such, transfer learning models can be used for both feature generation and classification.…”

Section: Methodsmentioning

confidence: 99%

Hybrid Deep Feature Generation for Appropriate Face Mask Use Detection

Aydemir

Yalçınkaya

Barua

et al. 2022

IJERPH

View full text Add to dashboard Cite

Mask usage is one of the most important precautions to limit the spread of COVID-19. Therefore, hygiene rules enforce the correct use of face coverings. Automated mask usage classification might be used to improve compliance monitoring. This study deals with the problem of inappropriate mask use. To address that problem, 2075 face mask usage images were collected. The individual images were labeled as either mask, no masked, or improper mask. Based on these labels, the following three cases were created: Case 1: mask versus no mask versus improper mask, Case 2: mask versus no mask + improper mask, and Case 3: mask versus no mask. This data was used to train and test a hybrid deep feature-based masked face classification model. The presented method comprises of three primary stages: (i) pre-trained ResNet101 and DenseNet201 were used as feature generators; each of these generators extracted 1000 features from an image; (ii) the most discriminative features were selected using an improved RelieF selector; and (iii) the chosen features were used to train and test a support vector machine classifier. That resulting model attained 95.95%, 97.49%, and 100.0% classification accuracy rates on Case 1, Case 2, and Case 3, respectively. Having achieved these high accuracy values indicates that the proposed model is fit for a practical trial to detect appropriate face mask use in real time.

show abstract

Section: Methodsmentioning

confidence: 99%

Hybrid Deep Feature Generation for Appropriate Face Mask Use Detection

Aydemir

Yalçınkaya

Barua

et al. 2022

IJERPH

View full text Add to dashboard Cite

show abstract

“…The Northeastern University Steel Surface Defects dataset [21] is an established dataset containing images of defects on hot rolled steel that can be used to quickly evaluate image classification models. This is covered in more detail in ref [22]. The images were pre-processed by applying contrast-limited adaptive histogram equalization to remove the effects of relative brightness and then resizing so the images could be passed through VGG16.…”

Section: Baseline Model: Computer Vision and Transfer Learningmentioning

confidence: 99%

Neural message passing for predicting abnormal grain growth in Monte Carlo simulations of microstructural evolution

Cohn¹,

Holm²

2021

Preprint

Self Cite

View full text Add to dashboard Cite

Abnormal grain growth can significantly alter the properties of materials during processing. This can cause significant variation in the properties and performance of in-spec feedstock components subjected to identical processing paths. Understanding and controlling abnormal grain growth has proved to be elusive due to the stochastic nature of this phenomenon. However, recent advances in deep learning provide a promising alternative to traditional experimental and physics-based methods for understanding this phenomenon. Neural message passing allows deep learning to be applied to irregular inputs including graph representations of grain structures in a material. In this study we generate a large database of Monte Carlo simulations of abnormal grain growth in an idealized system. We apply message passing neural networks to predict the occurrence of abnormal grain growth in these simulations using only the initial state of the system as input. A computer vision model is also trained for the same task for comparison. The preliminary results indicate that the message passing approach outperforms the computer vision method and achieved 75% prediction accuracy, significantly better than random guessing. Analysis of the uncertainty in the Monte Carlo simulations provides a road map for ongoing work on this project.

show abstract

“…The optimal image technique successfully detects the presence or lack of fusion defects (created artificially) in powder bed fusion parts [67]. Machine learning techniques can successfully detect and classify the six surface defects in hot-rolled steel parts based on images [68]. The computer vision (CV) technique has been applied to capture the signs of microstructural features and classify them automatically into groups with high accuracy using relatively small data sets [69][70][71].…”

Section: Introductionmentioning

confidence: 99%

Image Processing of Mg-Al-Sn Alloy Microstructures for Determining Phase Ratios and Grain Size and Correction with Manual Measurement

et al. 2021

View full text Add to dashboard Cite

The study of microstructures for the accurate control of material properties is of industrial relevance. Identification and characterization of microstructural properties by manual measurement are often slow, labour intensive, and have a lack of repeatability. In the present work, the intermetallic phase ratio and grain size in the microstructure of known Mg-Sn-Al alloys were measured by computer vision (CV) technology. New Mg (Magnesium) alloys with different alloying element contents were selected as the work materials. Mg alloys (Mg-Al-Sn) were produced using the hot-pressing powder metallurgy technique. The alloys were sintered at 620 °C under 50 MPa pressure in an argon gas atmosphere. Scanning electron microscopy (SEM) images were taken for all the fabricated alloys (three alloys: Mg-7Al-5Sn, Mg-8Al-5Sn, Mg-9Al-5Sn). From the SEM images, the grain size was counted manually and automatically with the application of CV technology. The obtained results were evaluated by correcting automated grain counting procedures with manual measurements. The accuracy of the automated counting technique for determining the grain count exceeded 92% compared to the manual counting procedure. In addition, ASTM (American Society for Testing and Materials) grain sizes were accurately calculated (approximately 99% accuracy) according to the determined grain counts in the SEM images. Hence, a successful approach was proposed by calculating the ASTM grain sizes of each alloy with respect to manual and automated counting methods. The intermetallic phases (Mg17Al12 and Mg2Sn) were also detected by theoretical calculations and automated measurements. The accuracy of automated measurements for Mg17Al12 and Mg2Sn intermetallic phases were over 95% and 97%, respectively. The proposed automatic image processing technique can be used as a tool to track and analyse the grain and intermetallic phases of the microstructure of other alloys such as AZ31 and AZ91 magnesium alloys, aluminium, titanium, and Co alloys.

show abstract

Unsupervised Machine Learning Via Transfer Learning and k-Means Clustering to Classify Materials Image Data

Cited by 50 publications

References 34 publications

Hybrid Deep Feature Generation for Appropriate Face Mask Use Detection

Hybrid Deep Feature Generation for Appropriate Face Mask Use Detection

Neural message passing for predicting abnormal grain growth in Monte Carlo simulations of microstructural evolution

Image Processing of Mg-Al-Sn Alloy Microstructures for Determining Phase Ratios and Grain Size and Correction with Manual Measurement

Contact Info

Product

Resources

About