Article
Computer Science, Artificial Intelligence
S. R. Sannasi Chakravarthy, Harikumar Rajaguru
Summary: This paper proposes a computer-assisted approach for the effective classification of breast cancer severity, using digital mammograms and ensemble-based algorithms. The novelty lies in the successful implementation of a two-level cascaded classifier, which outperforms other ensemble classifiers in terms of classification performance.
Article
Chemistry, Multidisciplinary
Antoine Badi Mame, Jules-Raymond Tapamo
Summary: Facial Expression Recognition (FER) is a rapidly growing field with diverse applications. This study analyzes the performance of various local descriptors and classifiers in the FER problem and identifies the best combinations. The results show that conventional FER approaches are still comparable to state-of-the-art deep learning methods.
APPLIED SCIENCES-BASEL
(2022)
Article
Automation & Control Systems
Zuowei Zhang, Songtao Ye, Yiru Zhang, Weiping Ding, Hao Wang
Summary: This paper proposes a new classifier method based on evidence theory to handle missing values, improving classification performance by addressing the uncertainty and imprecision brought by incompleteness. Experimental results demonstrate that the proposed method outperforms other methods in terms of accuracy, precision, recall, and F1 measure, but with higher computational costs.
IEEE-CAA JOURNAL OF AUTOMATICA SINICA
(2022)
Article
Multidisciplinary Sciences
Lin Song, Huixuan Zhao, Zongfang Ma, Qi Song
Summary: This study proposes a two-level fusion-based method for the classification of construction waste. The method utilizes statistical histograms and gradient histograms, encodes and fuses visual features using the bag-of-visual-words method, and achieves automatic classification of construction waste through a joint decision-making model.
Article
Geography, Physical
Panfei Fang, Guanglong Ou, Ruonan Li, Leiguang Wang, Weiheng Xu, Qinling Dai, Xin Huang
Summary: This study proposes a new machine learning workflow to address the challenges in accurately obtaining a stand tree species map in large mountainous areas. It divides the study area into floristic regions, collects multiple data sets, and uses decision fusion strategies to improve classification accuracy. The method achieves an overall accuracy of 72.18% on the validation dataset and significantly improves accuracy compared to base classifiers.
GISCIENCE & REMOTE SENSING
(2023)
Article
Agriculture, Multidisciplinary
Jianlei Kong, Hongxing Wang, Xiaoyi Wang, Xuebo Jin, Xing Fang, Seng Lin
Summary: The MCF-Net model proposed in this paper combines the requirements for identifying different crop species from actual farmland scenes, and utilizes the cross-stage partial network (CSPNet) as the backbone module, three parallel sub-networks, and a cross-level fusion module. By leveraging massive fine-granulometric information, MCF-Net has superior representation ability for distinguishing interclass discrepancies and tolerating intra-class variances.
COMPUTERS AND ELECTRONICS IN AGRICULTURE
(2021)
Article
Physics, Multidisciplinary
Dongxue Zhao, Xin Wang, Yashuang Mu, Lidong Wang
Summary: The study reviewed the latest ensemble classification algorithms for dealing with imbalanced datasets, and experimental results showed that classical algorithms with a dynamic selection strategy provide a practical way to improve classification performance for both binary class and multi-class imbalanced datasets.
Article
Multidisciplinary Sciences
Lidia Garrido-Sanz, Miquel angel Senar, Josep Pinol
Summary: This study focuses on reducing false positive species caused by classifiers, and benchmarks two popular classifiers for accurate identification of target species. The intersection approach effectively reduces the number of false positive species when combining the results of the two classifiers. In addition, applying an analytical detection limit further decreases the number of false positive species.
Article
Environmental Sciences
Shan He, Peng Peng, Yiyun Chen, Xiaomi Wang
Summary: This paper investigates the optimal combination of feature selection methods and classifiers for crop classification. It constructs 18 multi-crop classification models and evaluates their performance. The results show that different feature selection methods have different effects on different models, and the classification strategy combining spectral, textual, and environmental indexes can improve crop recognition ability.
Article
Environmental Sciences
Weihua Chen, Jie Pan, Yulin Sun
Summary: This study explores the fusion method of GF-5 and Sentinel-2A images and achieves tree species classification using a random forest classifier. The results show that the fused image has higher spatial integration and spectral fidelity, and the classification accuracy is significantly improved.
Article
Chemistry, Analytical
Mohammad Al-Qaderi, Elfituri Lahamer, Ahmad Rad
Summary: A new architecture is proposed to address speaker identification challenges, utilizing prosodic and short-term spectral features for gender and identity classification. Experimental evaluations show promising results in improving recognition performance in challenging environments with low signal-to-noise ratio and short utterances.
Article
Remote Sensing
Ning Ye, Justin Morgenroth, Cong Xu, Na Chen
Summary: Understanding the composition and dynamics of New Zealand's woody vegetation communities is important for management. Integration of Sentinel-2 and PlanetScope imagery enabled cost-effective and accurate forest mapping. Different combinations of classifiers and imagery resulted in classification accuracies ranging from 0.669 to 0.956, with the best accuracy achieved through the integration of Sentinel-2 and PlanetScope imagery. Digital terrain model was identified as the most important feature for all scenarios.
INTERNATIONAL JOURNAL OF APPLIED EARTH OBSERVATION AND GEOINFORMATION
(2021)
Article
Automation & Control Systems
Jan Rabcan, Vitaly Levashenko, Elena Zaitseva, Miroslav Kvassay
Summary: This article discusses a fuzzy classifier-based approach for EEG signal classification. The results of the study indicate that fuzzy classifiers are effective tools for EEG signal classification and achieve the highest classification accuracy.
IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS
(2022)
Article
Biodiversity Conservation
Yasong Guo, Yinyi Lin, Wendy Y. Chen, Jing Ling, Qiaosi Li, Joseph Michalski, Hongsheng Zhang
Summary: This study developed a model to estimate aboveground biomass (AGB) in urban parks at the species level by combining remote sensing data, field data, and tree species identification methods. The model achieved high accuracy in estimating AGB and identified the impact of a typhoon on AGB in the study area. The study contributes to understanding AGB distribution and carbon cycle dynamics in urban scenarios.
ECOLOGICAL INDICATORS
(2022)
Review
Environmental Sciences
Maja Michalowska, Jacek Rapinski
Summary: Remote sensing techniques, especially Light Detection and Ranging (LiDAR), have greatly improved large-scale forest inventory by providing three-dimensional point cloud data for object extraction and classification. Various LiDAR-derived metrics, combined with classification algorithms, contribute to high accuracy in tree species discrimination. Full-waveform data extraction and the use of random forest or support vector machine classifiers have shown to be most effective in increasing species discrimination performance.
Article
Forestry
Joielan Xipaia dos Santos, Helena Cristina Vieira, Deivison Venicio Souza, Marlon Costa de Menezes, Graciela Ines Bolzon de Muniz, Patricia Soffiatti, Silvana Nisgoski
Summary: The integration of near infrared spectroscopy (NIR) with machine learning techniques is an effective method for discriminating wood species with commercial value. Analysis of Louros wood samples from the Brazilian Amazon using NIR and machine learning shows that discriminative patterns can be obtained from near infrared spectra across different anatomical sections, with excellent accuracy and F1-Scores obtained using the PLS-DA algorithm.
EUROPEAN JOURNAL OF WOOD AND WOOD PRODUCTS
(2021)
Article
Forestry
Thais A. P. Goncalves, Alexandre G. Navarro, Silvana Nisgoski, Julia Sonsin-Oliveira
Summary: Research conducted in the Encantado archaeological site in Maranhao state, Brazil, revealed that waterlogged wood pillars were mainly constructed using Tabebuia/Handroanthus trees, commonly known as ipe. By utilizing near-infrared spectroscopy (NIR) and principal component analysis (PCA), it was possible to differentiate the wood samples with approximately 80% variation between data, although the PCA did not separate them according to their types. Comparing the NIR spectra of wood pillars with recently sawn ipe wood helped to identify that the latter cannot be used to identify waterlogged wood.
WOOD SCIENCE AND TECHNOLOGY
(2021)
Article
Forestry
Rangel Consalter, Antonio Carlos Vargas Motta, Julierme Zimmer Barbosa, Fabiane Machado Vezzani, Rafael Alejandro Rubilar, Stephen A. Prior, Silvana Nisgoski, Marcos Vinicius Martins Bassaco
Summary: Research in southern Brazil found that mid-rotation application of N and P significantly increased commercial volume of Pinus taeda, while K, lime, and micronutrient applications had no noticeable effects. Nutrient and lime applications increased total litter accumulation, with K omission leading to an increase in total root mass.
EUROPEAN JOURNAL OF FOREST RESEARCH
(2021)
Article
Forestry
Silvana Nisgoski, Thais A. P. Goncalves, Julia Sonsin-Oliveira, Adriano W. Ballarin, Graciela I. B. Muniz
Summary: By analyzing near-infrared spectroscopy, different types of charcoal can be distinguished, aiding regulatory agencies in ensuring the sustainability of charcoal supply. The application of new technologies such as NIR for charcoal identification may improve the efficiency of government agents.
Article
Agriculture, Multidisciplinary
Claudio Leones Bazzi, Michel Rosin Martins, Bruno Eduardo Cordeiro, Luciano Gebler, Eduardo Godoy de Souza, Kelyn Schenatto, Pedro Luiz de Paula Filho, Ricardo Sobjak
Summary: Yield mapping technologies can improve both the quantity and quality of agricultural production, particularly in perennial crops. The system developed includes hardware and software components to quantify and relate quality to harvest decisions. Collaboration with research institutions and testing in apple orchards in southern Brazil shows potential for positive impact on the fruit sector.
PRECISION AGRICULTURE
(2022)
Article
Materials Science, Paper & Wood
Joielan Xipaia dos Santos, Helena Cristina Vieira, Tawani Lorena Naide, Deivison Venicio Souza, Graciela Ines Bolzon de Muniz, Patricia Soffiatti, Silvana Nisgoski
Summary: This study aimed to assess the potential of colorimetry in characterizing six species of the Fabaceae family, with P. suaveolens standing out in the Principal Component Analysis. The colorimetric parameters L*, b* and C* were found to be more important in the color results of the majority of wood samples, indicating the potential of colorimetry spectroscopy in characterizing Fabaceae species.
INTERNATIONAL WOOD PRODUCTS JOURNAL
(2021)
Article
Materials Science, Paper & Wood
Helena Cristina Vieira, Joielan Xipaia dos Santos, Deivison Venicio Souza, Polliana D'Angelo Rios, Graciela Ines Bolzon de Muniz, Simone Ribeiro Morrone, Silvana Nisgoski
Summary: This study evaluated the potential of colorimetry for differentiation of three species of the Fabaceae family in the Araucaria Forest in southern Brazil. The hue angle (h) showed the highest potential for species discrimination, while the a* and h parameters did not differ significantly among different wood samples.
MADERAS-CIENCIA Y TECNOLOGIA
(2022)
Proceedings Paper
Computer Science, Artificial Intelligence
Rafael Augusto de Oliveira, Michel Hanzen Scheeren, Pedro Joao Soares Rodrigues, Arnaldo Candido Junior, Pedro Luiz de Paula Filho
Summary: Face recognition is a challenging task, especially under adverse imaging conditions. Super-resolution techniques improve image quality and accuracy, with Generative Adversarial Networks being the state-of-the-art. In this study, a joint-learning approach was used to train a super resolution face recognition model, but the face recognition model did not converge.
OPTIMIZATION, LEARNING ALGORITHMS AND APPLICATIONS, OL2A 2022
(2022)
Proceedings Paper
Computer Science, Artificial Intelligence
Marcela Marques Barbosa, Everton Schneider dos Santos, Joao Paulo Teixeira, Saraspathy Naidoo Terroso Gama de Mendonca, Arnaldo Candido Junior, Pedro Luiz de Paula Filho
Summary: This study developed a software that can automatically count microbial colonies in Petri dishes and validated its efficiency through comparisons with manual counting results. The results showed a global correlation of 0.948 and an individual correlation of 0.8134 with manual counting. Therefore, it can be concluded that microbial counts can be performed automatically and reliably with the developed software.
OPTIMIZATION, LEARNING ALGORITHMS AND APPLICATIONS, OL2A 2022
(2022)
Article
Multidisciplinary Sciences
Joielan Xipaia dos Santos, Helena Cristina Vieira, Deivison Venicio Souza, Graciela Ines Bolzon De Muniz, Patricia Soffiatti, Silvana Nisgoski
Summary: The aim of this study was to verify the potential of colorimetric technique in identifying species marketed as tauari in the Brazilian Amazon. Parameters CIE L* a* b* were used to determine the color of wood samples from the State of Para and scientific collections. Results showed that parameter b* played a significant role in differentiating tauari samples based on color, especially in tangential and radial sections. PCA analysis revealed distinct color patterns between different sources of wood samples, with h and L* parameters providing valuable information for species identification. Therefore, colorimetric technique can be used as an auxiliary tool for wood identification, but it is recommended to be used in combination with anatomical description due to the complexity of species-level separation in the tauari group.
ANAIS DA ACADEMIA BRASILEIRA DE CIENCIAS
(2022)
Article
Forestry
Joielan-Xipaia Santos, Helena-Cristina Vieira, Deivison-Venicio Souza, Paulo-Afonso B. Costa, Graciela-Ines B. Muniz, Patricia Sofatti, Silvana Nisgoski
Summary: The study aimed to evaluate the potential of colorimetry for discriminating wood from the louros group of Brazilian native species. The results suggest that longitudinal surfaces are more suitable for characterizing this group.
Article
Materials Science, Paper & Wood
Elaine Cristina Lengowski, Eraldo Antonio Bonfatti Junior, Silvana Nisgoski, Graciela Ines Bolzon de Muniz, Umberto Klock
Summary: Thermal treatment significantly affects the anatomical structure, chemical composition, physical properties, mechanical properties, and color of teakwood, leading to improved thermal stability but decreased mechanical properties as well as a darker color.
MADERAS-CIENCIA Y TECNOLOGIA
(2021)
Article
Materials Science, Paper & Wood
Elaine Cristina Lengowski, Eraldo Antonio Bonfatti Junior, Rafael Dallo, Silvana Nisgoski, Jorge Luis Monteiro de Mattos, Jose Guilherme Prata
Summary: This study investigated the effects of adding nanocellulose to phenol-formaldehyde adhesive on the physico-mechanical properties of plywood panels, finding that NFC significantly influenced the adhesive properties, especially in mechanical tests.
MADERAS-CIENCIA Y TECNOLOGIA
(2021)
Article
Ecology
Angela Maria Stupp, Helena Cristina Vieira, Polliana D'Angelo Rios, Graciela Ines Bolzon de Muniz, Silvana Nisgoski
Summary: This study aimed to measure and compare anatomical elements of wood and charcoal of three species to support identification of seized materials. Changes in cell dimensions and behavior were observed after carbonization in the Fabaceae species evaluated.
Article
Materials Science, Paper & Wood
Helena Cristina Vieira, Joielan Xipaia dos Santos, Eliane Lopes da Silva, Polliana D'Angelo Rios, Graciela Ines Bolzon de Muniz, Simone Ribeiro Morrone, Silvana Nisgoski
Summary: This study aimed to evaluate the potential of near-infrared spectroscopy for discrimination of wood and charcoal from four Myrtaceae species native of the Araucaria Forest. Results showed that it was possible to separate different species based on anatomical sections and position in trunk for wood, but some more sample dispersion were observed for charcoal. Linear discriminant analysis corroborated the results.
WOOD MATERIAL SCIENCE & ENGINEERING
(2021)