Article
Multidisciplinary Sciences
Yuexing Han, Ruiqi Li, Shen Yang, Qiaochuan Chen, Bing Wang, Yi Liu
Summary: The analysis of complex microstructure images of materials currently relies on human experience and lacks automatic quantitative methods. This study proposes a machine learning-based center-environment segmentation (CES) feature model for image segmentation, which improves accuracy through the iterative introduction of domain knowledge and environment features.
SCIENTIFIC REPORTS
(2022)
Article
Geochemistry & Geophysics
Wenkang Zhang, Dan Liu, Chunjing Wang, Ruitao Liu, Daqian Wang, Longzhou Yu, Shuming Wen
Summary: This paper presents an improved algorithm for mineral flotation foam image segmentation in order to optimize the flotation process and improve the recovery of mineral resources. Python libraries are used for image enhancement and compensation, quantitative analysis of factors affecting image segmentation accuracy, and suggestions for improvement. The study analyzes the characteristics of the bubbles and the influence of flotation conditions on the foam image. An improved version of the watershed segmentation algorithm is used for segmentation analysis, resulting in higher accuracy and shorter segmentation time compared to the standard algorithm.
Article
Remote Sensing
Vahid Nasiri, Pawel Hawrylo, Piotr Janiec, Jaros law Socha
Summary: This study investigates the use of PlanetScope satellite images and pixel-based and object-based image analysis for accurate mapping of forest cover and detection of tree cuttings. Machine learning models were trained and evaluated, and the results showed that the object-based random forest classifier performed the best in both forest cover mapping and tree-cutting detection.
INTERNATIONAL JOURNAL OF APPLIED EARTH OBSERVATION AND GEOINFORMATION
(2023)
Article
Remote Sensing
Bingxiao Wu, Zhujun Gu, Wuming Zhang, Qinghua Fu, Maimai Zeng, Aiguang Li
Summary: This study proposes the investigator accuracy (IA) metric for image segmentation validation, focusing on the location accuracy of single patches. It evaluates the capture accuracy of near-center subregions and category weight to determine segmentation quality. Grayscale dilation and erosion algorithms are optimized, and a parallel analysis scheme is applied for efficient IA evaluation. Results show that the capture accuracy and category weight of a patch affect its IA.
INTERNATIONAL JOURNAL OF APPLIED EARTH OBSERVATION AND GEOINFORMATION
(2023)
Article
Computer Science, Artificial Intelligence
Serban Oprisescu, Radu-Mihai Coliban, Mihai Ivanovici
Summary: Texture characterization is a valuable tool for analyzing object surface images in various fields. This paper proposes a new hand-crafted texture characterization technique based on light polarization property. The technique utilizes a circular polarization filter in the image acquisition process to capture polarization signatures that can locally characterize texture. Experimental results demonstrate the usefulness of the proposed method for surface/material classification and color image segmentation.
PATTERN RECOGNITION LETTERS
(2022)
Article
Mining & Mineral Processing
David G. Shatwell, Victor Murray, Augusto Barton
Summary: Sensor-based ore sorting is a technology that classifies high-grade mineralized rocks from low-grade waste rocks to reduce operation costs. This paper presents an ore-sorting algorithm based on image processing and machine learning that can classify rocks from a gold and silver mine based on their grade. The proposed algorithm achieves a Matthews correlation coefficient of 0.961 points and a processing time under 44 ms, promising for real-time ore sorting applications.
INTERNATIONAL JOURNAL OF MINING SCIENCE AND TECHNOLOGY
(2023)
Article
Biochemical Research Methods
Changye Yang, Sriram Baireddy, Valerian Meline, Enyu Cai, Denise Caldwell, Anjali S. Iyer-Pascuzzi, Edward J. Delp
Summary: In this study, a set of metrics to quantify plant wilting was developed and tested. These metrics can be used to identify wilting caused by different stresses in different plant species, and are important for studying plant resistance genes and genomic regions.
Article
Computer Science, Artificial Intelligence
Julian Luengo, Raul Moreno, Ivan Sevillano, David Charte, Adrian Pelaez-Vegas, Marta Fernandez-Moreno, Pablo Mesejo, Francisco Herrera
Summary: This paper reviews and categorizes computer vision techniques for metallographic image segmentation, introduces deep learning-based ensemble techniques utilizing pixel similarity, and conducts thorough comparisons in real-world datasets to discuss strengths, weaknesses, and application frameworks. The paper also addresses open challenges in the field to provide guidance for future research to fill existing gaps.
INFORMATION FUSION
(2022)
Article
Chemistry, Analytical
Soumaya Dghim, Carlos M. Travieso-Gonzalez, Radim Burget
Summary: This paper introduces the detection of Nosema disease using image processing tools, machine learning, and deep learning approaches. Two main strategies are examined: one involves extracting valuable information and features from microscopic images dataset using image processing tools and applying machine learning methods, while the other explores deep learning and transfer learning.
Review
Dentistry, Oral Surgery & Medicine
Lang Zhang, Wang Li, Jinxun Lv, Jiajie Xu, Hengyu Zhou, Gen Li, Keqi Ai
Summary: This article reviews recent advances in computer-aided segmentation methods for oral and maxillofacial surgery and discusses their advantages and limitations. The study found that these methods can be divided into traditional image processing and machine learning categories, with machine learning methods showing unprecedented performance. However, challenges such as scarcity of datasets and visible artifacts in images still exist. Accurate image segmentation is crucial for precise treatment and surgical planning in oral and maxillofacial surgery.
JOURNAL OF DENTISTRY
(2023)
Article
Computer Science, Information Systems
Bernat Galmes, Gabriel Moya-Alcover, Pedro Bibiloni, Javier Varona, Antoni Jaume-i-Capo
Summary: This article presents a robust segmentation method for measuring toenails. The method is used in a clinical trial to objectively quantify the incidence of a specific pathology. It uses the Hough transform to locate the tip of the toe and estimate the nail location and size, and then classifies the super-pixels based on their geometric and photometric information. The watershed transform is then used to delineate the border of the nail.
MULTIMEDIA TOOLS AND APPLICATIONS
(2022)
Article
Agriculture, Multidisciplinary
Liang Gong, Xiaofeng Du, Kai Zhu, Chenghui Lin, Ke Lin, Tao Wang, Qiaojun Lou, Zheng Yuan, Guoqiang Huang, Chengliang Liu
Summary: The study of plant growth state relies on root architecture parameters, with root segmentation being crucial to measuring these parameters. A new method based on a convolutional neural network was proposed for pixel-level segmentation of rice roots under strong noise, achieving an intersection over union (IoU) of 87.4%. This approach provides an automatic and fast pixel-level root segmentation method, essential for root morphology analysis.
COMPUTERS AND ELECTRONICS IN AGRICULTURE
(2021)
Article
Computer Science, Artificial Intelligence
Zhonghao Zhang, Qiqiang Li, Wen Song, Pengfei Wei, Jing Guo
Summary: In this paper, a novel automatic cell segment method based on concave point detection and matching is proposed. By using a deep neural network and high-quality image deblurring techniques, the method can accurately extract cell contours and address the challenges of segmenting touching cells by detecting and matching concave points.
EXPERT SYSTEMS WITH APPLICATIONS
(2022)
Article
Computer Science, Artificial Intelligence
Shouvik Chakraborty, Kalyani Mali
Summary: The absence of dedicated vaccines or drugs makes COVID-19 a global pandemic, and early diagnosis is identified as an effective prevention mechanism. A novel unsupervised machine learning method called SUFMACS is proposed for efficiently interpreting and segmenting COVID-19 radiological images. The results demonstrate the efficiency and real-life applicability of this approach in investigating both CT scan and X-ray images.
EXPERT SYSTEMS WITH APPLICATIONS
(2021)
Article
Computer Science, Information Systems
Katiuscia Mannaro, Matteo Baire, Alessandro Fanti, Matteo Bruno Lodi, Luca Didaci, Alessandro Fedeli, Luisanna Cocco, Andrea Randazzo, Giuseppe Mazzarella, Giorgio Fumera
Summary: This paper addresses the problem of automatic image segmentation methods applied to the production process of traditional Sardinian flatbread. A machine learning algorithm based on support vector machines is proposed for the segmentation and measurement estimation of bread images. Experimental results demonstrate the accuracy and efficiency of the method in accurately segmenting bread sheet images and extracting representative dimensions.
Article
Mathematics
Wan-Yu Tsai
Summary: This paper investigates the set of genuine small representations of the nonlinear double cover similar to G, and proves that similar to s./2 (similar to G) is precisely the set of genuine irreducible representations arising from the KazhdanPatterson lifting of the trivial representation when similar to G is simply laced and split.
INTERNATIONAL MATHEMATICS RESEARCH NOTICES
(2023)
Review
Physics, Applied
Albina Jetybayeva, Nikolay Borodinov, Anton V. Ievlev, Md Inzamam Ul Haque, Jacob Hinkle, William A. Lamberti, J. Carson Meredith, David Abmayr, Olga S. Ovchinnikova
Summary: Imaging mass spectrometry (IMS) is a powerful analytical technique used in various fields that provides qualitative compositional analysis and spatial mapping. The combination of mass spectra with spatial information produces large high-dimensional datasets, making automated computational methods crucial for exploratory analysis.
JOURNAL OF APPLIED PHYSICS
(2023)
Article
Biochemical Research Methods
Md Inzamam Ul Haque, Debangshu Mukherjee, Sylwia A. Stopka, Nathalie Y. R. Agar, Jacob Hinkle, Olga S. Ovchinnikova
Summary: This study correlates H&E-stained biopsy data with MALDI mass-spectrometric imaging data to determine cancerous regions and their unique chemical signatures. The correlation between optical H&E features and chemical information in MSI allows for the prediction of prostate cancer with around 80% accuracy. Additionally, two chemical biomarkers were found to predict cancerous regions.
JOURNAL OF THE AMERICAN SOCIETY FOR MASS SPECTROMETRY
(2023)
Article
Chemistry, Physical
Jonghee Yang, Diana K. LaFollette, Benjamin J. Lawrie, Anton V. Ievlev, Yongtao Liu, Kyle P. Kelley, Sergei V. Kalinin, Juan-Pablo Correa-Baena, Mahshid Ahmadi
Summary: Mixed cesium- and formamidinium-based metal halide perovskites (MHPs) are promising photovoltaic materials, but high cesium ratios result in chemical complexities and local inhomogeneities, compromising the optoelectronic performance.
ADVANCED ENERGY MATERIALS
(2023)
Article
Nanoscience & Nanotechnology
Wan-Yu Tsai, Shelby B. Pillai, Karthik Ganeshan, Saeed Saeed, Yawei Gao, Adri C. T. van Duin, Veronica Augustyn, Nina Balke
Summary: In this study, the deformation of birnessite (& delta;-MnO2) electrode during charge storage in aqueous electrolytes was investigated. The effect of electrolyte cation and electrode morphology on the deformation was analyzed using operando atomic force microscopy (AFM) and molecular dynamics (MD) simulation. The results showed that the electrode underwent expansion during cation intercalation in both K2SO4 and Li2SO4 electrolytes, but with different potential dependencies. The stronger cation-birnessite interaction in Li2SO4 electrolyte resulted in higher local stress heterogeneity, leading to pronounced electrode degradation.
ACS APPLIED MATERIALS & INTERFACES
(2023)
Article
Chemistry, Multidisciplinary
Yongtao Liu, Jonghee Yang, Benjamin J. Lawrie, Kyle P. Kelley, Maxim Ziatdinov, Sergei V. Kalinin, Mahshid Ahmadi
Summary: The increasing photovoltaic efficiency and stability of metal halide perovskites (MHPs) are attributed to the improvement in understanding the microstructure of polycrystalline MHP thin films. A workflow combining conductive atomic force microscopy (AFM) measurement with a machine learning (ML) algorithm was designed to systematically investigate the grain boundaries in MHPs. This approach revealed that the properties of grain boundaries play critical roles in MHP stability.
Article
Chemistry, Physical
Yongtao Liu, Jonghee Yang, Rama K. Vasudevan, Kyle P. Kelley, Maxim Ziatdinov, Sergei Kalinin, Mahshid Ahmadi
Summary: We demonstrate an active machine learning framework for driving an automated scanning probe microscope (SPM) to discover the microstructures responsible for specific aspects of transport behavior in metal halide perovskites (MHPs). This approach allows the microscope to discover the microstructural elements that maximize the onset of conduction, hysteresis, or any other characteristic derived from a set of current-voltage spectra. It provides new opportunities for exploring the origins of materials functionality in complex materials by SPM and can be integrated with other characterization techniques.
JOURNAL OF PHYSICAL CHEMISTRY LETTERS
(2023)
Article
Chemistry, Physical
Yongtao Liu, Rama K. K. Vasudevan, Kyle P. Kelley, Hiroshi Funakubo, Maxim Ziatdinov, Sergei V. V. Kalinin
Summary: We developed automated experiment workflows for identifying the best predictive channel in spectroscopic measurements. The approach combines ensembled deep kernel learning for probabilistic predictions and reinforcement learning for channel selection. The implementation in multimodal imaging of piezoresponse force microscopy (PFM) showed that the amplitude is the best predictive channel for polarization-voltage and frequency-voltage hysteresis loop areas. This workflow and code can be applied to other multimodal imaging and local characterization methods.
NPJ COMPUTATIONAL MATERIALS
(2023)
Article
Computer Science, Artificial Intelligence
Yongtao Liu, Anna N. Morozovska, Eugene A. Eliseev, Kyle P. Kelley, Rama Vasudevan, Maxim Ziatdinov, Sergei V. Kalinin
Summary: Using hypothesis-learning-driven automated scanning probe microscopy (SPM), this study investigates the bias-induced transformations in various devices and materials. It is crucial to understand these mechanisms on the nanometer scale with a wide range of control parameters, which is experimentally challenging. The hypothesis-driven SPM autonomously identifies the mechanisms of bias-induced domain switching and reveals the importance of kinetic control.
Article
Computer Science, Artificial Intelligence
Arpan Biswas, Rama Vasudevan, Maxim Ziatdinov, Sergei Kalinin
Summary: Unsupervised and semi-supervised ML methods like VAE are widely used in physics, chemistry, and materials sciences for disentangling representations and finding latent manifolds in complex experimental data. This study explores a latent Bayesian optimization approach for hyperparameter trajectory optimization in unsupervised and semi-supervised ML, demonstrated by joint-VAE with rotational invariances. The method is applied to finding joint discrete and continuous rotationally invariant representations in the MNIST database and a plasmonic nanoparticles material system.
MACHINE LEARNING-SCIENCE AND TECHNOLOGY
(2023)
Article
Physics, Applied
Jinyuan Yao, Yongtao Liu, Shaoqing Ding, Yanglin Zhu, Zhiqiang Mao, Sergei V. Kalinin, Ying Liu
Summary: Ferroelectricity in van der Waals layered material has attracted significant attention. The ferroelectric properties of CuInP2S6 (CIPS), which is the only van der Waals layered material that has demonstrated ferroelectricity in the bulk, have been observed to persist even at a few nanometers thickness. However, the potential device applications of CIPS' ferroelectric properties are just beginning to be explored.
APPLIED PHYSICS LETTERS
(2023)
Article
Computer Science, Artificial Intelligence
Maxim Ziatdinov, Chun Yin (Tommy) Wong, Sergei Kalinin
Summary: Recent advances in scanning tunneling and transmission electron microscopies have generated large volumes of imaging data containing information on the structure and functionality of materials. However, automatic extraction and classification of patterns in the images is non-trivial. To address this problem, the authors propose a shift-invariant variational autoencoder approach and demonstrate its effectiveness on 1D, synthetic, and experimental data. The shift VAE analysis shows promise for pattern discovery, but also has limitations.
MACHINE LEARNING-SCIENCE AND TECHNOLOGY
(2023)
Article
Nanoscience & Nanotechnology
Wan-Yu Tsai, Shelby B. B. Pillai, Karthik Ganeshan, Saeed Saeed, Yawei Gao, Adri C. T. van Duin, Veronica Augustyn, Nina Balke
Summary: The effect of electrolyte cation and electrode morphology on the deformation of birnessite (delta-MnO2) during charge storage in aqueous electrolytes was investigated. The results showed that the delta-MnO2 electrode underwent expansion during cation intercalation, but with different potential dependencies. The stronger cation-birnessite interaction in Li2SO4 electrolyte led to higher local stress heterogeneity, which might be the reason for the pronounced electrode degradation in this electrolyte.
ACS APPLIED MATERIALS & INTERFACES
(2023)
Article
Automation & Control Systems
Sheryl Sanchez, Yongtao Liu, Jonghee Yang, Sergei V. Kalinin, Maxim Ziatdinov, Mahshid Ahmadi
Summary: In recent years, laboratory automation and high-throughput synthesis and characterization have become increasingly important in the research community. To effectively analyze the large datasets and extract system properties, suitable machine learning techniques, such as the variational autoencoder (VAE) approach, are needed. This study explores the binary library of metal halide perovskite microcrystals using low-dimensional latent representations of photoluminescence spectra. The combination of translationally invariant variational autoencoders (tVAEs) and conditional autoencoders (cVAEs) allows for a deeper understanding of the underlying mechanisms within the data.
ADVANCED INTELLIGENT SYSTEMS
(2023)