Percorrer por autor "Candido Junior, Arnaldo"
A mostrar 1 - 10 de 11
Resultados por página
Opções de ordenação
- Assessment of honey bee cells using deep learningPublication . Alves, Thiago da Silva; Ventura, Paulo J.C.; Neves, Cátia J.; Candido Junior, Arnaldo; Paula Filho, Pedro L. de; Pinto, M. Alice; Rodrigues, Pedro JoãoTemporal assessment of honey bee colony strength is required for different applications in many research projects. This task often requires counting the number of cells with brood and food reserves multiple times a year from images taken in the apiary. There are thousands of cells in each frame, which makes manual counting a time-consuming and tedious activity. Thus, the assessment of frames has been frequently been performed in the apiary in an approximate way by using methods such as the Liebefeld. The automation of this process using modern imaging processing techniques represents a major advance. The objective of this work was to develop a software capable of extracting each cell from frame images, classify its content and display the results to the researcher in a simple way. The cells’ contents display a high variation of patterns which added to light variation make their classification by software a challenging endeavor. To address this challenge, we used Deep Neural Networks (DNNs) for image processing. DNNs are known by achieving the state-of-art in many fields of study including image classification, because they can learn features that best describe the content being classified, such as the interior of frame cells. Our DNN model was trained with over 60,000 manually labeled images whose cells were classified into seven classes: egg, larvae, capped larvae, honey, nectar, pollen, and empty. Our contribution is an end-to-end software capable of doing automatic background removal, cell detection, and classification of its content based on an input image. With this software the researcher is able to achieve an average accuracy of 94% over all classes and get better results compared with approximation methods and previous techniques that used handmade features like color and texture.
- Automatic detection and classification of honey bee comb cells using deep learningPublication . Alves, Thiago da Silva; Pinto, M. Alice; Ventura, Paulo J.C.; Neves, Cátia J.; Biron, David G.; Candido Junior, Arnaldo; Paula Filho, Pedro L. de; Rodrigues, Pedro JoãoIn a scenario of worldwide honey bee decline, assessing colony strength is becoming increasingly important for sustainable beekeeping. Temporal counts of number of comb cells with brood and food reserves offers researchers data for multiple applications, such as modelling colony dynamics, and beekeepers information on colony strength, an indicator of colony health and honey yield. Counting cells manually in comb images is labour intensive, tedious, and prone to error. Herein, we developed a free software, named DeepBee©, capable of automatically detecting cells in comb images and classifying their contents into seven classes. By distinguishing cells occupied by eggs, larvae, capped brood, pollen, nectar, honey, and other, DeepBee© allows an unprecedented level of accuracy in cell classification. Using Circle Hough Transform and the semantic segmentation technique, we obtained a cell detection rate of 98.7%, which is 16.2% higher than the best result found in the literature. For classification of comb cells, we trained and evaluated thirteen different convolutional neural network (CNN) architectures, including: DenseNet (121, 169 and 201); InceptionResNetV2; InceptionV3; MobileNet; MobileNetV2; NasNet; NasNetMobile; ResNet50; VGG (16 and 19) and Xception. MobileNet revealed to be the best compromise between training cost, with ~9 s for processing all cells in a comb image, and accuracy, with an F1-Score of 94.3%. We show the technical details to build a complete pipeline for classifying and counting comb cells and we made the CNN models, source code, and datasets publicly available. With this effort, we hope to have expanded the frontier of apicultural precision analysis by providing a tool with high performance and source codes to foster improvement by third parties (https://github.com/AvsThiago/DeepBeesource).
- Classificação do conteúdo de favos em quadros de colmeias usando Deep LearningPublication . Alves, Thiago da Silva; Ventura, Paulo J.C.; Neves, Cátia J.; Pinto, M. Alice; Candido Junior, Arnaldo; Paula Filho, Pedro L. de; Rodrigues, Pedro JoãoNo âmbito de várias tarefas da investigação apícola, existe uma que obriga o investigador a classificar e contar o conteúdo de cada favo em cada quadro da colmeia. Esta tarefa tem por objetivo analisar e controlar a progressão da criação, de abelhas, e de reservas, o que implica repeti-la múltiplas vezes a cada ano. Cada quadro contém milhares de favos o que leva a que a contagem, na maior parte dos casos, seja feita de forma aproximada. Os favos podem conter: pupas (criação fechada), larvas em diferentes fases de maturação, mel, néctar, pólen, ovos, ou então podem estar vazios. A automatização deste processo, com o auxílio de um sistema computacional, representa uma importante evolução na referida tarefa. O presente trabalho aborda a classificação automática do conteúdo de favos a partir de imagens digitais. Arquiteturas neuronais de Deep Learning têm mostrado um bom potencial a classificar padrões que exibem elevada variabilidade visual. Assim, a utilização deste método de aprendizagem máquina adequa-se à complexidade e variabilidade visual dos padrões apresentados pelas imagens dos favos. No modelo desenvolvido neste trabalho foi utilizada a arquitetura neuronal GoogleNet. Esta foi treinada utilizando 63344 imagens anotadas e separadas nas sete classes referidas. A taxa média de acerto do modelo sobre o conjunto de validação foi de 94% o que melhora substancialmente o resultado obtido com técnicas clássicas (SVM - 76%). Este estudo foi financiado pelo projeto BEEHOPE através do concurso conjunto 2013-2014 BiodivErsA/FACCE-JPI pela FCT (Portugal), CNRS (França) e MEC (Espanha).
- Classification of control/pathologic subjects with support vector machinesPublication . Teixeira, Felipe; Fernandes, Joana Filipa Teixeira; Guedes, Victor; Candido Junior, Arnaldo; Teixeira, João PauloThe diagnosis of pathologies using vocal acoustic analysis has the advantage of been noninvasive and inexpensive technique compared to traditional technique in use. In this work the SVM were experimentally tested to diagnose dysphonia, chronic laryngitis or vocal cords paralysis. Three groups of parameters were experimented. Jitter, shimmer and HNR, MFCCs extracted from a sustained vowels and MFCC extracted from a short sentence. The first group showed their importance in this type of diagnose and the second group showed low discriminative power. The SVM functions and methods were also experimented using the dataset with and without gender separation. The best accuracy was 71% using the jitter, shimmer and HNR parameters without gender separation.
- Determination of harmonic parameters in pathological voices-efficient algorithmPublication . Fernandes, Joana Filipa Teixeira; Freitas, Diamantino Silva; Candido Junior, Arnaldo; Teixeira, João PauloFeatured Application The paper describes a low-complexity/efficient algorithm to determine the short-term Autocorrelation, HNR, and NHR in sustained vowel audios, to be used in stand-alone devices with low computational power. These parameters can be used as input features of a smart medical decision support system for speech pathology diagnosis. The harmonic parameters Autocorrelation, Harmonic to Noise Ratio (HNR), and Noise to Harmonic Ratio are related to vocal quality, providing alternative measures of the harmonic energy of a speech signal. They will be used as input resources for an intelligent medical decision support system for the diagnosis of speech pathology. An efficient algorithm is important when implementing it on low-power devices. This article presents an algorithm that determines these parameters by optimizing the window type and length. The method used comparatively analyzes the values of the algorithm, with different combinations of window and size and a reference value. Hamming, Hanning, and Blackman windows with lengths of 3, 6, 12, and 24 glottal cycles and various sampling frequencies were investigated. As a result, we present an efficient algorithm that determines the parameters using the Hanning window with a length of six glottal cycles. The mean difference of Autocorrelation is less than 0.004, and that of HNR is less than 0.42 dB. In conclusion, this algorithm allows extraction of the parameters close to the reference values. In Autocorrelation, there are no significant effects of sampling frequency. However, it should be used cautiously for HNR with lower sampling rates.
- Harmonic to noise ratio measurement - selection of window and lengthPublication . Fernandes, Joana Filipa Teixeira; Teixeira, Felipe; Guedes, Victor; Candido Junior, Arnaldo; Teixeira, João PauloHarmonic to Noise Ratio (HNR) measures the ratio between periodic and non-periodic components of a speech sound. It has become more and more important in the vocal acoustic analysis to diagnose pathologic voices. The measure of this parameter can be done with Praat software that is commonly accept by the scientific community has an accurate measure. Anyhow, this measure is dependent with the type of window used and its length. In this paper an analysis of the influence of the window and its length was made. The Hanning, Hamming and Blackman windows and the lengths between 6 and 24 glottal periods were experimented. Speech files of control subjects and pathologic subjects were used. The results showed that the Hanning window with the length of 12 glottal periods gives measures of HNR more close to the Praat measures.
- Image processing of petri dishes for counting microorganismsPublication . Barbosa, Marcela; Santos, Everton; Teixeira, João Paulo; Mendonca, Saraspathy; Candido Junior, Arnaldo; Paula Filho, Pedro Luiz deFor a food to be considered functional it is necessary to prove that it has microorganisms in its composition. In order to determine the presence of microorganisms in a food, laboratory analyzes can be carried out using of Petri dishes, which must pass through an incubation period, to then manually count the number of bacterial colonies that are present in each board. Therefore, the objective of this work was to develop a software to perform the automated counting of colonies contained in Petri plates and to validate the efficiency of the tool through comparisons with the results of a manual count. For this, an algorithm was developed based on digital image processing techniques capable of identifying the Petri dish within the image and counting the number of colonies present on each plate. As a result, a global correlation of 0.948 was observed in relation to the manual count, and in an individual analysis, a correlation of 0.8134 was obtained. Thus, it can be concluded that counts of microorganisms can be performed automatically and reliably with the developed software.
- Leaf-based species recognition using convolutional neural networksPublication . Pires, Willian Oliveira; Fernandes, Ricardo Corso; Paula Filho, Pedro Luiz de; Candido Junior, Arnaldo; Teixeira, João PauloIdentifying plant species is an important activity in specie control and preservation. The identification process is carried out mainly by botanists, consisting of a comparison of already known specimens or using the aid of books, manuals or identification keys. Artificial Neural Networks have been shown to perform well in classification problems and are a suitable approach for species identification. This work uses Convolutional Neural Networks to classify tree species by leaf images. In total, 29 species were collected. This work analyzed two network models, Darknet-19 and GoogLeNet (Inception-v3), presenting a comparison between them. The Darknet and GoogLeNet models achieved recognition rates of 86.2% and 90.3%, respectively.
- Long short term memory on chronic laryngitis classificationPublication . Guedes, Victor; Candido Junior, Arnaldo; Fernandes, Joana Filipa Teixeira; Teixeira, Felipe; Teixeira, João PauloThe classification study with the use of machine learning concepts has been applied for years, and one of the aspects in which this can be applied is for the analysis of speech acoustics applied to the analysis of pathologies. Among the pathologies present, one of them is chronic laryngitis. Thus, this article aims to present the results for a classification of chronic laryngitis with the use of Long Short Term Memory as a classifier. The parameters of relative jitter, relative shimmer and autocorrelation was used as input of the LSTM. A dataset of about 1500 instances were used to train, validate and test along 4 experiments with LSTM and one feedforward Artificial Neural Network (ANN). The results of the LSTM overcome the ones of the feedforward ANN, and was about 100% accuracy, sensitivity and specificity in test set, denoting a promising future for this classification tool in the voice pathologies diagnose.
- Smart data driven system for pathological voices classificationPublication . Fernandes, Joana; Candido Junior, Arnaldo; Freitas, Diamantino Silva; Teixeira, João PauloClassifying and recognizing voice pathologies non-invasively using acoustic analysis saves patient and specialist time and can improve the accuracy of assessments. In this work, we intend to understand which models provide better accuracy rates in the distinction between healthy and pathological, to later be implemented in a system for the detection of vocal pathologies. 194 control subjects and 350 pathological subjects distributed across 17 pathologies were used. Each subject has 3 vowels in 3 tones, which is equivalent to 9 sound files per subject. For each sound file, 13 parameters were extracted (jitta, jitter, Rap, PPQ5, ShdB, Shim, APQ3, APQ5, F0, HNR, autocorrelation, Shannon entropy and logarithmic entropy). For the classification between healthy and pathological, several classifiers were used (Decision Trees, Discriminant Analysis, Logistic Regression Classifiers, Naive Bayes Classifiers, Support Vector Machines, Nearest Neighbor Classifiers, Ensemble Classifiers, Neural Network Classifiers) with various models. For each patient, 118 parameters were used (13 acoustic parameters * 9 sound files per subject, plus the subject's gender). As pre-processing of the input matrix data, the Outliers treatment was used using the quartile method, then the data were normalized and, finally, Principal Component Analysis (PCA) was applied in order to reduce the dimension. As the best model, the Wide Neural Network was obtained, with an accuracy of 98% and AUC of 0.99.