Search

The use of machine learning techniques to predict yield based on remote sensing is a no-return path and studies conducted on farm aim to help rural producers in decision-making. Thus, commercial fields equipped with technologies in Mato Grosso, Brazil, were monitored by satellite images to predict cotton yield using supervised learning techniques. The objective of this research was to identify how early in the growing season, which vegetation indices and which machine learning algorithms are best to predict cotton yield at the farm level. For that, we went through the following steps: 1) We observed the yield in 398 ha (3 fields) and eight vegetation indices (VI) were calculated on five dates during the growing season. 2) Scenarios were created to facilitate the analysis and interpretation of results: Scenario 1: All Data (8 indices on 5 dates = 40 inputs) and Scenario 2: best variable selected by Stepwise regression (1 input). 3) In the search for the best algorithm, hyperparameter adjustments, calibrations and tests using machine learning were performed to predict yield and performances were evaluated. Scenario 1 had the best metrics in all fields of study, and the Multilayer Perceptron (MLP) and Random Forest (RF) algorithms showed the best performances with adjusted R2 of 47% and RMSE of only 0.24 t ha-1, however, in this scenario all predictive inputs that were generated throughout the growing season (approx. 180 days) are needed, so we optimized the prediction and tested only the best VI in each field, and found that among the eight VIs, the Simple Ratio (SR), driven by the K-Nearest Neighbor (KNN) algorithm predicts with 0.26 and 0.28 t ha-1 of RMSE and 5.20% MAPE, anticipating the cotton yield with low error by ±143 days, and with important aspect of requiring less computational demand in the generation of the prediction when compared to MLP and RF, for example, enabling its use as a technique that helps predict cotton yield, resulting in time savings for planning, whether in marketing or in crop management strategies.

Made available in DSpace on 2022-05-01T10:18:56Z (GMT). No. of bitstreams: 0 Previous issue date: 2021-12-01 Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES) Single-target regression can accurately predict the crop's performance but fails to generalize problems with more than one true and cross-validatable solution. An alternative to output multiple numeric values upon the input, we think, would be multi-target regression (MTR) with either Random Forest (RF) or k-nearest neighbors (KNN). Therefore, we captured the advantages of high-resolution remote sensing and multi-target machine learning into an immersive single framework then analyzed if it could be possible for accurately predicting for the yield of sweet potato by the market class of its tuberous roots (i.e., Extra < 0.15 kg; 015 ≤ Extra AA ≤ 0.45 kg; and Extra A > 0.45 kg) upon imagery data on summer and winter full-scale fields. The remote sensing captured the spectral changes on both fields and enabled the MTR to accurately predict for the yield of sweet potato in total and by the market class of harvestable roots upon normalized difference vegetation index (NDVI) and its derivative version (GreenNDVI) as well as upon soil-adjusted vegetation index (SAVI). The SAVI-RF framework predicted the summer field to yield marketable roots at the proportions of 2.04 t ha−1 Extra, 3.89 t ha−1 Extra AA and 2.08 t ha−1 Extra A, and the spectral data from the mid-stage of cultivation at 296 growing degree days (GDD) minimized its mean absolute error (MAE) to 2.66 t ha−1. The GNDVI-RF framework predicted the winter field to yield 1.64 t ha−1 Extra, 5.02 t ha−1 Extra AA and 3.65 t ha−1 Extra A, with an error of 3.45 t ha−1 upon spectral data from sampling on the late stage at 966 GDD. Our insights are timely an absolutely will open up the horizons for harvesting high-quality roots to commercialization, industrialization and propagation, and scaling up this essentially provocative yet emerging crop for food safety and energy security. Department of Engineering and Mathematical Sciences São Paulo State University Department of Computing School of Sciences São Paulo State University Department of Engineering and Mathematical Sciences São Paulo State University Department of Computing School of Sciences São Paulo State University CAPES: 001

Made available in DSpace on 2022-05-01T05:29:32Z (GMT). No. of bitstreams: 0 Previous issue date: 2021-09-01 Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES) Sweet potato is a tuberous root with versatility in food products, but also with applications in the energy industry, such as in ethanol production. Developing mechanisms to assess the performance of this crop is important, difficult, and costly, as its commercial product grows below ground. The use of remote sensing to evaluate the development of sweet potato has not yet been reported in the literature. In our study, we showed that spectral vegetation indices are good proxies to monitor the temporal dynamics of crop growth and differentiate phenological stages, regardless of the growing season. The development phases were divided into three stages according to the vegetation indices: (I) initial stage ( 500 GDD), when major changes in VIs no longer occur because vegetative growth has ceased. Besides that, we found that these indices can predict crop yield before harvest. In two growing seasons, the smallest errors in yield estimates occurred during the growth stage. In the summer season with NDVI at 355 GDD with errors of 2.63 t ha−1 and in the winter season when GNDVI at 440 GDD had errors of 3.06 t ha−1. Department of Engineering and Mathematical Sciences São Paulo State University Department of Agriculture Federal University Lavras Department of Agronomy Taquaritinguense Institute of Higher Education Department of Engineering and Mathematical Sciences São Paulo State University

Agriculture 4.0 involves the incorporation of information and communication technologies into machines, equipment, and sensors for use in agricultural production systems. It aims to ease decision-making in agricultural processes. Statistical Quality Control (SQC) is a statistical method with several techniques and tools used to analyze the variability. These tools can be used to provide important information for decision making, including for mechanized agricultural operations. This paper aimed to characterize the worldwide scientific literature on Statistical Process Control use in mechanized agricultural processes, demonstrating its potential to be incorporated into Agriculture 4.0. Our research involved a bibliometric survey on Scopus and Academic Google databases. The analyzed studies allowed us to infer that SQC tools may improve understanding of mechanized operations and be used in Agriculture 4.0. Such features can also streamline and enhance decision making, converting big data into useful information. RESUMO A Agricultura 4.0 envolve a incorporação de tecnologias de informação e comunicação em máquinas, equipamentos e sensores para uso em sistemas de produção agrícola e tem como principal objetivo, facilitar a tomada de decisões em processos agrícolas. O Controle Estatístico de Qualidade (CEQ) é um método estatístico que utiliza várias técnicas e ferramentas para analisar a variabilidade dos dados, ferramentas estas que podem ser utilizadas para fornecer informações importantes para a tomada de decisões, inclusive para operações agrícolas mecanizadas. Este trabalho teve como objetivo buscar na literatura científica mundial o estado da arte na utilização do Controle Estatístico de Qualidade em processos agrícolas mecanizados, demonstrando também o seu potencial para ser incorporado na Agricultura 4.0. Nossa pesquisa envolveu um levantamento bibliométrico nas bases de dados Scopus e Google Acadêmico. Os estudos analisados permitiram inferir que as ferramentas CEQ podem melhorar a compreensão das operações mecanizadas e serem utilizadas na Agricultura 4.0. Esses recursos também podem agilizar e aprimorar a tomada de decisões, de modo a converter dados massivos (big data) em informações úteis e precisas (right data).

Made available in DSpace on 2021-06-25T11:13:27Z (GMT). No. of bitstreams: 0 Previous issue date: 2021-01-01 One of the main problems in the peanut production process is to identify the pod maturity stage. Peanut plants have indeterminate growth, which leads to a high pod maturity variability within the same plant. Moreover, the actual method of determining maturity is destructive and highly subjectivity, which does not represent the overall variability in the field. Hence, the main goal of this study was to verify the possibility to estimate peanut maturity and its in-field variability using an alternative non-destructive method based on orbital remote sensing. High-resolution satellite images (~ 3 m) were obtained from the PlanetScope platform for two commercial peanut fields in São Paulo state, Brazil, during the reproductive stage of the peanut crop (89 to 118 days after sowing—DAS). The fields were divided into 54 plots (30 × 30 m). The maturity was obtained using the Hull Scrape method. All Vegetation Indices (VIs) used showed a high Pearson correlation (p < 0.001) between peanut maturity and the VIs, with values decreasing as maturity increased. Non-Linear Index (NLI) values from 0.561 to 0.465 suggested that pods reached greater maturity than 74% (inflection point). The results found in this study indicated a great potential to use high-resolution satellite images to predict peanut maturity variability in commercial field. In addition, the proposed method contributes to monitoring the dynamics spatio-temporal of maturity progression, allowing for more accurate in-season and inversion management strategies in peanut. Department of Agriculture Lavras Federal University (UFLA), Aquenta Sol Department of Engineering and Exact Sciences São Paulo State University (UNESP), Via Access Prof. Paulo Donato Castellane s/n Department of Crop and Soil Sciences University of Georgia, Tifton Campus, 2360 Rainwater Road Department of Engineering and Exact Sciences São Paulo State University (UNESP), Via Access Prof. Paulo Donato Castellane s/n

Made available in DSpace on 2020-12-12T02:36:38Z (GMT). No. of bitstreams: 0 Previous issue date: 2020-04-01 Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES) One way to improve the quality of mechanized cotton harvesting is to change harvester settings and adjustments throughout the process, according to information obtained during the operation. We believe that yield predictions are important for managing the quality of operation, aiming at increasing efficiency and reducing losses. Therefore, this study aimed to develop an automated system for cotton yield prediction from color images acquired by a simple mobile device. We propose a robust approach to environmental conditions, training detection algorithms with images acquired at different times throughout the day, and evaluating three different scenarios (low-, average-, and high-demand computational resources). The experimental results for the average demand computational scenario, which are suitable for real-time deployment on low-cost devices such as smartphones and other ARM-processed devices, indicated the possibility of counting bolls using images acquired at different times throughout the day, with mean errors of 8.84% (∼5 bolls). Furthermore, we observed a 17.86% error when predicting yield using 205 images from the testing dataset, which is equivalent to about 19.14 g. São Paulo State University School of Agricultural and Veterinary Sciences (UNESP/FCAV) São Paulo State University School of Agricultural and Veterinary Sciences (UNESP/FCAV)

Submitted by Danilo Tedesco de Oliveira (danilo.tedesco@unesp.br) on 2019-04-24T19:58:51Z No. of bitstreams: 1 Dissertação_Danilo_Tedesco.pdf: 1445381 bytes, checksum: 6f583321b60356da1da852c9e78746e5 (MD5) Rejected by Neli Silvia Pereira null (nelisps@fcav.unesp.br), reason: Solicitamos que realize correções na submissão seguindo as orientações abaixo: 1 - O agradecimento à CAPES deve ter a seguinte redação: O presente trabalho foi realizado com apoio da Coordenação de Aperfeiçoamento de Pessoal de Nível Superior - Brasil (CAPES) - Código de Financiamento 001 Abaixo segue a portaria que regulamenta. 2 - Falta o abstract (resumo em inglês) no repositório. 3 - Na ficha catalográfica, a entrada está por Oliveira e no repositório por Tedesco Oliveira. As entradas não podem ser diferentes, ou é uma, ou outra. Agradecemos a compreensão. PORTARIA Nº 206, DE 4 DE SETEMBRO DE 2018 Dispõe sobre obrigatoriedade de citação da CAPES O PRESIDENTE DA COORDENAÇÃO DE APERFEIÇOAMENTO DE PESSOAL DE NÍVEL SUPERIOR, no uso das atribuições que lhe foram conferidas pelo art. 26 do (a) Estatuto, aprovado (a) pelo Decreto nº 8977, de 30/01/2017, e CONSIDERANDO o indicado nos Editais da CAPES, nos Termos de Compromisso de Bolsista, nos regulamentos de bolsas no exterior e de bolsas no país, no Manual de AUXPE, e no termo de adesão ao Portal de Periódicos; CONSIDERANDO o constante dos autos do processo nº 23038.013648/2018-51, resolve: Art. 1º Os trabalhos produzidos ou publicados, em qualquer mídia, que decorram de atividades financiadas, integral ou parcialmente, pela CAPES, deverão, obrigatoriamente, fazer referência ao apoio recebido. Art. 2º Para fins de identificação da fonte de financiamento fica autorizada a utilização do código 001 para todos os financiamentos recebidos. Art. 3º Deverão ser usadas as seguintes expressões, no idioma do trabalho: "O presente trabalho foi realizado com apoio da Coordenação de Aperfeiçoamento de Pessoal de Nível Superior - Brasil (CAPES) - Código de Financiamento 001 "This study was financed in part by the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior - Brasil (CAPES) - Finance Code 001" Art. 4º Fica o pró-reitor de pós-graduação ou congênere, responsável pela divulgação e aplicação da regra dentro das Instituições de Ensino Superior que recebem apoio da CAPES. Art. 5º A falha em obedecer esta norma implicará em mudanças eventuais nos apoios da CAPES para as instituições e pesquisadores envolvidos, a partir de 2020. Art. 6º Esta Portaria entra em vigor na data de sua publicação. ABILIO A. BAETA NEVES Fonte: CAPES Diário Oficial da União: http://pesquisa.in.gov.br/imprensa/jsp/visualiza/index.jsp?data=05/09/2018&jornal=515&pagina=22 on 2019-04-25T12:58:19Z (GMT) Submitted by Danilo Tedesco de Oliveira (danilo.tedesco@unesp.br) on 2019-04-25T13:33:42Z No. of bitstreams: 1 Dissertação_Danilo_Tedesco.pdf: 1477295 bytes, checksum: 45f613d2d6f945498beedac60cad5ddf (MD5) Approved for entry into archive by Tatiana Camila Gricio (tatiana.gricio@unesp.br) on 2019-04-25T18:12:33Z (GMT) No. of bitstreams: 1 oliveira_dt_me_jabo.pdf: 1477295 bytes, checksum: 45f613d2d6f945498beedac60cad5ddf (MD5) Made available in DSpace on 2019-04-25T18:12:33Z (GMT). No. of bitstreams: 1 oliveira_dt_me_jabo.pdf: 1477295 bytes, checksum: 45f613d2d6f945498beedac60cad5ddf (MD5) Previous issue date: 2019-03-01 Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES) Uma maneira de melhorar a qualidade da operação de colheita mecanizada do algodão é alterar as configurações e regulagens de colhedora ao longo do processo de acordo com informações que podem ser adquiridas durante a própria colheita. Acreditamos que a previsão da produtividade pode se tornar uma informação de grande importância no gerenciamento da qualidade da operação, visando ao aumento da eficiência e redução das perdas no processo. Portanto, nessa pesquisa apresentamos o desenvolvimento de um sistema inteligente capaz de prever a produtividade do algodão em imagens coloridas, adquiridas por um dispositivo mobile simples. Propomos uma abordagem robusta às condições ambientais, treinando algoritmos de detecção com imagens adquiridas em diferentes horários ao longo do dia. Os resultados experimentais para o modelo SSD Mobilenet V1 indicaram que é possível realizar a contagem dos capulhos presentes nas imagens adquiridas em diferentes horários ao longo do dia, com erros médios de 8.84% (~5 capulhos). Além disso, avaliando a previsão da produtividade em 205 imagens do conjunto de dados de teste, o erro foi igual à 17.86%. Esse valor representa variações de até 19.14 gramas entre cada previsão. One way to improve the quality of the mechanized cotton harvesting operation is to change settings and adjustments of the harvester throughout the process according to information acquired during the operation. We believe that yield prediction can become important information in managing the quality of the operation aiming at increasing efficiency and reducing losses in the process. Therefore, we present in this research the development of an intelligent system capable of predicting cotton yield from color images acquired by a simple mobile device. We propose a robust approach to environmental conditions, training detection algorithms with images acquired at different times throughout the day. The experimental results for the SSD MobileNet V1 model indicated the possibility of counting bolls present in the images acquired at different times throughout the day, with mean errors of 8.84% (~5 bolls). Also, the error was equal to 17.86% when evaluating yield prediction on 205 images from the test dataset. This value represents variations of up to 19.14 grams between each prediction. Código de financiamento 001

Among the factors that may influence corn yield are sowing operation direction and plant population, since they alter the shade index between crop rows, and consequently the efficiency of interception of the photosynthetically active radiation by the leaves of plants. The aim of this study was to evaluate the influence of sowing direction and plant populations on P2830VYH hybrid corn yield. The experiment was conducted at the facilities of FCAV-UNESP, latitude 21º 14’ S and longitude 48º 16’ W in Jaboticabal (SP). The treatments were composed of three sowing directions (North-South, Northeast-Southwest, and East-West) and three population densities (40,000, 60,000 and 80,000 plants ha-1). The treatment with 80,000 plants ha-1 had the lowest cost per sack of corn and profitability 21% higher than that of the treatment with 60,000 plants ha-1, and 45% higher than that of the treatment with 40,000 plants ha-1. Sowing direction had no influence on grain yield when environmental conditions were favorable to the development of the crop. Lower plant densities lead to higher individual results in relation to the number of cobs, length of cobs and number of grains per row. RESUMO Entre os fatores que podem influenciar a produtividade do milho estão o sentido da operação de semeadura e a população de plantas, pois esses alteram o índice de sombreamento entre as linhas da cultura, e consequentemente a eficiência da interceptação da radiação fotossinteticamente ativa pelas folhas das plantas. Teve-se por objetivo nesse estudo avaliar a influência do sentido de semeadura e populações de plantas sobre a produtividade do híbrido de milho P2830VYH. O experimento foi conduzido em área da FCAV–UNESP, à latitude de 21º 14’ S e longitude 48º 16’ W no município de Jaboticabal (SP). Os tratamentos estudados foram compostos por três direções de semeadura (Norte-Sul; Nordeste-Sudoeste; e Leste-Oeste) e três densidades populacionais (40.000; 60.000; e 80.000 plantas ha-1). A lavoura com 80.000 plantas ha-1 apresentou o menor custo por saca de milho, e lucratividade 21% maior que o tratamento de 60.000 plantas ha-1, e 45% maior que o tratamento de 40.000 plantas ha-1. O sentido da semeadura não exerceu influência sobre a produtividade de grãos quando as condições ambientais foram favoráveis ao desenvolvimento da lavoura. Menores densidades de plantas apresentam maiores resultados individuais em relação ao número de espigas, comprimento de espigas e número de grãos por fileira.

The variation in population density in a corn crop can positively, negatively or neutrally affects plant productivity depending on the productive potential of the area. The aim of this work was to evaluate the effects of the variation in corn seed dosage on crop yield, and define from which percentage of variation the productivity of the sown line is affected negatively. The experiment was installed at FCAV-UNESP, in Jaboticabal (SP), Brazil. Twelve variations on plant population were evaluated, ranging from -27% to 27%, varying with a frequency of 4.5%. The morphological and productive characteristics of each treatment were evaluated through regression analysis. Each 1% of negative variation on seed dosage was lost 1.06% in corn yield. Positive variations, however, presented changes that were 0 to -2.59%. Negative variations on seed dosage reduced corn productivity by up to 28%. The positive variation affects the productivity of the crop in a less accentuated way, with a reduction in productivity that reaches 2.59% in the largest variations and may even cause positive productivity results depending on the maximum potential of the area.

The tools of precision agriculture are of utmost importance in the Brazilian agribusiness, enabling increases in yields and reducing production costs. The use of harvest monitoring systems makes it possible due the possibility to identify pontual problems in an area, however, it becomes necessary to be working properly so it does not acquire incorrect information. Therefore, the purpose with this study was to propose a new approach to identify discrepant points in harvesting maps using statistical process control, as well as to define the best multiple of the standard deviation to identificate these points. The work was conducted during the soybean harvesting at São Geronimo farm in an area of 38 hectares in the municipality of Candido Mota, located in the the state of São Paulo. For gathering information, it was used a Stara crop monitoring system (model Topper Maps) set to record information during harvest in each three second. The productivity data were used to generate an individual control chart to identify points that were out of control so they could be removed. Two standard deviation multiples, that presented an average productivity closer to the average real productivity of the area, were selected. The multiples of the deviations that came closest were the 2σ and 3σ. Two multiples of standard deviation presented an average yield closer to the average real yield of the area. Individual control charts can be used to set control limits and identify possible discrepancies. The multiple of standard deviation 3σ presented information with greater reliability.