Search

Delineation of site-specific nutrient management zones (MZ) provides a basis for practical and cost-effective management of spatial soil fertility in precision agriculture. Therefore, the objective...

The increasing market interest in coffee beverage, lead coffee growers around the world to adopt more efficient methods to select the best-quality coffee beans. Currently, coffee beans selection is carried out either manually, which is a costly and unreliable process, or using electronic sorting machines, which are often inefficient because some coffee beans defects, such as sour and immature beans, have similar spectral response patterns. In this sense, the present work aimed to assess coffee beans quality using both computer vision and machine learning techniques, such as Support Vector Machine (SVM), Deep Neural Network (DNN) and Random Forest (RF). For this purpose, an algorithm written in Python language was developed to extract shape and color features from coffee beans images. The obtained dataset was then used as input to the machine learning algorithms. The data reported in this study pointed to the importance of color descriptors for classifying coffee beans defects. Among the variables used, the components from RGB (Red, Green and Blue) and HSV (Hue, Saturation and Value) color spaces presented the most relevant contribution for the classification models. Also, the results reported in this study provides evidence that computer vision along with machine learning algorithms can be used to identify and classify coffee beans with a very high accuracy (> 90%). Key words: Deep neural network; classification; artificial intelligence; image processing; granulometry.

Optical sensors combined with machine learning algorithms have led to significant advances in seed science. These advances have facilitated the development of robust approaches, providing decision-making support in the seed industry related to the marketing of seed lots. In this study, a novel approach for seed quality classification is presented. We developed classifier models using Fourier transform near-infrared (FT-NIR) spectroscopy and X-ray imaging techniques to predict seed germination and vigor. A forage grass (Urochloa brizantha) was used as a model species. FT-NIR spectroscopy data and radiographic images were obtained from individual seeds, and the models were created based on the following algorithms: linear discriminant analysis (LDA), partial least squares discriminant analysis (PLS-DA), random forest (RF), naive Bayes (NB), and support vector machine with radial basis (SVM-r) kernel. In the germination prediction, the models individually reached an accuracy of 82% using FT-NIR data, and 90% using X-ray data. For seed vigor, the models achieved 61% and 68% accuracy using FT-NIR and X-ray data, respectively. Combining the FT-NIR and X-ray data, the performance of the classification model reached an accuracy of 85% to predict germination, and 62% for seed vigor. Overall, the models developed using both NIR spectra and X-ray imaging data in machine learning algorithms are efficient in quickly, non-destructively, and accurately identifying the capacity of seed to germinate. The use of X-ray data and the LDA algorithm showed great potential to be used as a viable alternative to assist in the quality classification of U. brizantha seeds.

This work aimed to determine potential areas for the establishment of cocoa moniliasis in Bahia state, Brazil, by means of fuzzy logic, based on historical datasets of temperature and air relative humidity, available for 519 measurement points distributed throughout the state of Bahia. The data were initially submitted to a descriptive statistical analysis. The spatial variability was determined through geostatistical analysis, followed by interpolation to map the spatial-temporal structure dependence of the phenomenon. Simulations of continuous pixel-to-pixel classification of variables were performed using fuzzy mapping to model the climatic risk of disease establishment. The exponential fuzzy model was applied to temperature data, while the linear model was used for air relative humidity data. The potential areas were defined for each month, using data of temperature and air relative humidity. The fuzzy models used allowed for modeling of the climatic risk of cocoa moniliasis establishment. A large area of the state is at high risk of disease, thus requiring mitigating measures to avoid the pathogen’s introduction and dissemination. RESUMO Com este trabalho objetivou-se definir áreas potenciais de estabelecimento da monilíase do cacaueiro no Estado da Bahia, por meio da lógica fuzzy, com base em séries históricas de temperatura e umidade relativa do ar, disponíveis para 519 pontos de medição distribuídos no Estado da Bahia. Os dados foram submetidos inicialmente a uma análise estatística descritiva. A variabilidade espacial foi determinada através de análise geoestatística, seguida de interpolações para mapear a estrutura da dependência espaço-temporal do fenômeno. Para modelar o risco climático de estabelecimento da doença foram realizadas simulações de classificação contínua, pixel-a-pixel, das representações das variáveis, utilizando o mapeamento fuzzy. Aos dados de temperatura, aplicou-se o modelo fuzzy exponencial, enquanto que aos dados de umidade relativa do ar, o modelo utilizado foi o linear. As definições de áreas potenciais foram realizadas para cada mês, utilizando a informação conjunta das pertinências da temperatura e da umidade relativa do ar. Os modelos fuzzy utilizados permitiram modelar o risco climático de estabelecimento da monilíase no Estado da Bahia. A maior extensão do Estado da Bahia tem riscos elevados para o desenvolvimento da doença, o que exige medidas mitigadoras para evitar a introdução e disseminação do patógeno.

ALGORITMO CLASSIFICADOR BAYESIANO ITERATIVO DE CODIGO ABERTO PARA AVALIACAO DA QUALIDADE DE GRAOS DE CAFE BENEFICIADOS A selecao de graos de cafe desempenha um papel fundamental na qualidade final do produto. Apos o processamento, os graos de cafe sao classificados de acordo com a quantidade de defeitos. Tradicionalmente, essa classificacao e executada manualmente, o que torna o processo trabalhoso e demorado. Este problema pode ser resolvido com tecnicas de processamento digital de imagens, uma vez que os graos defeituosos possuem caracteristicas visuais unicas. Considerando a dificuldade de classificacao manual dos defeitos, este trabalho teve como objetivo elaborar um algoritmo classificador bayesiano para identificar esses defeitos em graos de cafe beneficiados, com base em sua forma e cor. Para tal, foram utilizados 630 graos de cafe arabica, somando oito imagens ao todo. O algoritmo objetivou classificar quatro classes, que foram: graos normais, graos normais quebrados, graos pretos e graos pretos quebrados. Para avaliar a precisao do algoritmo classificador, calculou-se a exatidao global e o coeficiente Kappa, o que permite inferir se o classificador e melhor que uma classificacao aleatoria. Concluiu-se que o algoritmo desenvolvido apresentou uma precisao global de 76% e kappa igual a 0,6. Alem disso, a metodologia proposta mostrou grande potencial para aplicacao na avaliacao da qualidade de outros produtos, cujos parâmetros de forma e espectrais sao relevantes na avaliacao de sua qualidade.Palavras-chave: qualidade de graos de cafe; processamento digital de imagens; Jupyter Notebook; classificacao supervisionada. ABSTRACT: The selection of coffee beans plays a key role in the product's final quality. After processing, coffee beans are classified according to their quantity of defects. Traditionally this classification is performed manually, which makes the process laborious and time-consuming. This problem can be solved with digital image processing techniques since defective grains have unique visual characteristics. Considering the difficulty of manual classification of the defects, this study aimed to elaborate a Bayesian classifier algorithm to identify these defects in benefited coffee beans, based on its shape and color. To do so, 630 grains of arabica coffee were used, composing eight images in total. The algorithm aimed to classify four classes, which were: regular beans, normal broken beans, black beans, and black broken beans. In order to evaluate the accuracy of the classifier algorithm, it was calculated the global accuracy and the Kappa coefficient, which allows inferring if the classifier is better than a random classification. It was concluded that the developed algorithm presented a global accuracy of 76% and kappa equals to 0.6. Also, the proposed methodology showed great potential for application in the quality evaluation of other products, whose shape and spectral parameters are relevant in evaluating its quality.Keywords: coffee beans quality; digital image processing; Jupyter Notebook; supervised classification.

This study evaluated the use of low-cost materials for radiometric calibration of multispectral images. Four materials were tested: plywood panels painted with matte paint (M1); plywood panels cove...

Spatial interpolation methods are frequently used to characterize soil attributes’ spatial variability. However, inconclusive results, about the comparative performance of these methods, have been reported in the literature. Therefore, the present study aimed to analyze the efficiency of ordinary kriging (OK) and inverse distance weighting (IDW) methods in estimating the soil penetration resistance (SPR), soil bulk density (SBD), and soil moisture content (SM) using two distinct sampling grids. The soil sampling was performed on a 5.7 ha area in Southeast Brazil. For data collection, a regular grid with 145 points (20 x 20 m) was created. Soil samples were taken at a 0.20 m layer depth. In order to compare the accuracy of OK and IDW, another grid was created from the initial grid (A), by eliminating one interspersed line, which resulted in a grid with 41 sampled points (40 x 40 m). Results showed that sampling grid A presented less errors than B, proving that the more sampling points, the lower the errors that are associated with both methods will be. Overall, the OK was less biased than IDW only for SBD (A) and SM (B) maps, whereas IDW outperformed OK for the other attributes for both sampling grids.

Soil organic carbon (SOC) stocks are a remarkable property for soil and environmental monitoring. The understanding of their dynamics in crop soils must go forward. The objective of this study was to determine the impact of temporal environmental controlling factors obtained by satellite images over the SOC stocks along soil depth, using machine learning algorithms. The work was carried out in São Paulo state (Brazil) in an area of 2577 km2. We obtained a dataset of boreholes with soil analyses from topsoil to subsoil (0–100 cm). Additionally, remote sensing covariates (30 years of land use history, vegetation indexes), soil properties (i.e., clay, sand, mineralogy), soil types (classification), geology, climate and relief information were used. All covariates were confronted with SOC stocks contents, to identify their impact. Afterwards, the abilities of the predictive models were tested by splitting soil samples into two random groups (70 for training and 30% for model testing). We observed that the mean values of SOC stocks decreased by increasing the depth in all land use and land cover (LULC) historical classes. The results indicated that the random forest with recursive features elimination (RFE) was an accurate technique for predicting SOC stocks and finding controlling factors. We also found that the soil properties (especially clay and CEC), terrain attributes, geology, bioclimatic parameters and land use history were the most critical factors in controlling the SOC stocks in all LULC history and soil depths. We concluded that random forest coupled with RFE could be a functional approach to detect, map and monitor SOC stocks using environmental and remote sensing data.

Soil tillage systems alter soil physical attributes and may affect crop growth and yield. In this sense, the objective of this study was to assess the short-term impacts of tillage systems on soil physical properties, spectral response, and yield of the bean crop. The field experiment was laid out in a randomized block design with three tillage systems (NT: No-tillage; MT: Minimum tillage; and CT: Conventional tillage) and six replicates. Data collected included soil physical properties (SPR: Soil penetration resistance, SBD: bulk density, and SWC: water content), crop's spectral response (NDVI: Normalized difference vegetation index) through different multispectral sensors, and lastly, grain yield. Results showed that SBD and SPR values were significantly higher in the NT system at 9 days after planting. Moreover, the SPR in the NT system remained significantly higher (p

The spatiotemporal pattern of particulate matter (PM) concentrations is an important factor in predicting health issues in inhabitants of urban areas. The integration of satellite-derived aerosol optical depth (AOD) data with ground-level PM concentration data, obtained from monitoring networks, has contributed to better characterization of the spatiotemporal variability of aerosols worldwide. However, before using satellite AOD data as a proxy for PM in epidemiological and air quality studies in specific regions, the applicability of that strategy must be evaluated. In this study, we evaluate the use of the high-resolution AOD, derived from Multi-Angle Implementation of Atmospheric Correction (MAIAC) algorithm, as a predictor of surface PM concentrations in the Metropolitan Area of Sao Paulo (MASP). We found relatively weak or negative correlations between PM concentrations and MAIAC AOD, even after vertical correction by planetary boundary layer height and the hygroscopic growth factor. The weak correlations reported in this study are mainly due to the mismatch between the current MAIAC aerosol model and the properties of local aerosols in the MASP. Our results suggest that sources of aerosol particles in the MASP are quite diverse and that there is therefore no single optical model suitable for use with satellite-derived AOD.

Characterizing the land surface temperature (LST) and its diurnal cycle is important in understanding a range of surface properties, including soil moisture status, evaporative response, vegetation stress and ground heat flux. While remote-sensing platforms present a number of options to retrieve this variable, there are inevitable compromises between the resolvable spatial and temporal resolution. For instance, the spatial resolution of geostationary satellites, which can provide sub-hourly LST, is often too coarse (3 km) for many applications. On the other hand, higher-resolution polar orbiting satellites are generally infrequent in time, with return intervals on the order of weeks, limiting their capacity to capture surface dynamics. With recent developments in the application of unmanned aerial vehicles (UAVs), there is now the opportunity to collect LST measurements on demand and at ultra-high spatial resolution. Here, we detail the collection and analysis of a UAV-based LST dataset, with the purpose of examining the diurnal surface temperature response: something that has not been possible from traditional satellite platforms at these scales. Two separate campaigns were conducted over a bare desert surface in combination with either Rhodes grass or a recently harvested maize field. In both cases, thermal imagery was collected between 0800 and 1700 local solar time. The UAV-based diurnal cycle was consistent with ground-based measurements, with a mean correlation coefficient and root mean square error (RMSE) of 0.99 and 0.68 °C, respectively. LST retrieved over the grass surface presented the best results, with an RMSE of 0.45 °C compared to 0.67 °C for the single desert site and 1.28 °C for the recently harvested maize surface. Even considering the orders of magnitude difference in scale, an exploratory analysis comparing retrievals of the UAV-based diurnal cycle with METEOSAT geostationary data yielded pleasing results (R = 0.98; RMSE = 1.23 °C). Overall, our analysis revealed a diurnal range over the desert and maize surfaces of ~20 °C and ~17 °C respectively, while the grass showed a reduced amplitude of ~12 °C. Considerable heterogeneity was observed over the grass surface at the peak of the diurnal cycle, which was likely indicative of the varying crop water status. To our knowledge, this study presents the first spatially varying analysis of the diurnal LST captured at ultra-high resolution, from any remote platform. Our findings highlight the considerable potential to utilize UAV-based retrievals to enhance investigations across multi-disciplinary studies in agriculture, hydrology and land-atmosphere investigations.