Alaine Margarete Guimarães (Project Leader)
Eduardo Fávero Caires
José Carlos Ferreira da Rocha
Marcelo Giovanetti Canteri
Maria Salete Marcon Gomes Vaz
Selma Regina Aranha Ribeiro
Description:
In agriculture, there is a constant search for methods to improve productivity within the same volume of area, and various technologies are used for this purpose. Currently, the growing use of remote sensing by Unmanned Aerial Vehicles – UAVs (Unmanned Aerial Vehicle – UAV, Unmanned Aircraft Systems – UASs) can be highlighted. With the evolution and popularization of UAVs, several scientific and civil use projects have been developed, although in a modest manner in Brazil, due to the advantages of their use compared to other conventional methods. These advantages include cost and data collection time reduction, as well as increased safety and agility in this collection. Agriculture can benefit from aerial observation at all stages of production, with a focus on obtaining optical data. In this regard, different UAVs offer better design and performance compared to conventional photographic reconnaissance aircraft, as they are lightweight and small in size, have low flight speed (which implies better optical data acquisition), higher maximum altitude, and extreme resistance. The objective of this project is to deepen the study of UAV use in agriculture, with the development of innovative agricultural solutions, specifically in the acquisition and processing of agricultural data. Research Line: Computing, automation, and data management in agriculture.
Alaine Margarete Guimarães
Marcelo Giovanetti Canteri (Project Leader)Description:
In traditional trial analyses, for some variables, little numerical difference is observed between treatments, often leading to no significant statistical differences being found, but showing trends. This can result in subjective conclusions or the absence of conclusions, potentially leaving science at the mercy of marketing. Meta-analysis, among other advantages, allows us to avoid this lack of conclusions by performing a joint analysis of several previously published experiments. Meta-analysis is a systematic statistical synthesis of previous research results on a topic, emphasizing the production of quantitative conclusions. Therefore, it differs from a narrative review used in theses and dissertations. It allows statistically significant conclusions for variables that, under traditional experimentation in isolated trials, do not reach a level of significance. Data from published works, field records, or a mixture with new data collected by the meta-analysis executor can be used. The application of the technique has grown in the last decade and has already been used in more than 5,000 published works in the medical field, but it is still little used in the agronomic field. This proposal aims to: a) develop a computational system for performing meta-analysis aimed at agronomic experiments; and b) perform meta-analysis applied to phytopathological problems in four distinct areas, according to the specialties of the researchers involved in the project. The research areas involve topics where, in traditional trials, little numerical difference is observed between treatments, often leading to no significant statistical differences. The four areas addressed will be: no-tillage (straw, soil moisture, soil disturbance, and previous crop affecting diseases and productivity), soil fertility (nitrogen, phosphorus, and potassium affecting disease severity), seed treatment (chemical, biological, and physical affecting productivity), fungicide application methods (timing, rate, drop size, and adjuvant affecting disease control efficiency) for soybean, corn, wheat, and beans. Trials will be conducted for each topic over two agricultural years, and extensive bibliographic surveys will also be carried out using virtual means (journal portals) and physical means (libraries). Numerical data from each article will be extracted to generate tables. Meta-analysis will use these tables for new combined statistical calculations. At this stage, the best model for data analysis will be defined, among fixed effects models (FIXED), random effects models (RANDOM), maximum likelihood (ML), restricted maximum likelihood (REML), method of moments (MM), and multiple inputs with maximum likelihood (MLMI). The final phase will consist of software validation and the systematic and numerical summarization of the combined data, which will not be a simple narrative review. The executing team is active in the postgraduate program in Agronomy at UEL and in the postgraduate program in Applied Computing at UEPG and will have the support of researchers from EMBRAPA and UFRS, as well as the support of Prof. Dr. Laurence V. from Ohio State University. Research Line: Applied Computational Modeling.
Alceu de Souza Britto Jr. (Project Leader)
Luciano José Senger
Rosane Falate
Description:
A series of agricultural problems, such as plant disease detection, crop development analysis, grain selection, among others, can be overcome through image processing techniques. The objective of this research project is to develop and apply robust digital image processing techniques to agricultural problems to advance interdisciplinary research that can bring significant results to both computing and agriculture. Research Line: Computing, automation, and data management in agriculture.
Researchers: Alaine Margarete Guimarães (Project Leader)
Rafael Mazer Etto (Project Leader)Description:
During fruit ripening, physiological and biochemical changes occur in response to the differential expression of various genes. In climacteric fruits, such as tomatoes and bananas, ripening occurs concomitantly with a peak in respiration and an increase in ethylene production. In non-climacteric fruits, such as grapes and strawberries, there is no respiratory peak despite responding to exogenous ethylene. Studies conducted mainly on tomatoes, a climacteric pulp fruit model, have provided important information on the involvement of ethylene in the ripening of climacteric fruits; however, little is known about this process in non-climacteric fruits. Due to the large number of well-established genetic transformation techniques, available genetic and genomic tools, and the presence of varieties categorized as climacteric and non-climacteric, melon (Cucumis melo L.) has become an alternative model for studying pulp fruit ripening, ethylene perception, and signaling. Therefore, this work aims to identify the differentially expressed genes in climacteric melons of the Cantalupensis group (Cucumis melo var. cantalupensis) of the charentais type and non-climacteric melons of the inodorus group (Cucumis melo var. inodorus) of the valenciano type at different stages of fruit ripening. Techniques such as subtractive hybridization and qPCR will be used. To evaluate the influence of ethylene in the ripening process, fruits from the different study groups will be subjected to the presence and absence of this phytohormone application days after full flowering on the plant or after harvest. Thus, the aim is to identify important genes in the non-climacteric fruit ripening cascade and then quantify them to, if possible, propose a regulatory model for this group. Research Line: Applied computational modeling.
Rosane Falate (Project Leader)
Description: This project aims to develop a computational system for determining root characteristics, according to the needs identified by area specialists. Research Line: Computation, automation, and data management in agriculture.
Arion de Campos Junior
Jose Carlos Ferreira da Rocha
Rafael Mazer Etto (Project Leader)
The soil governs plant productivity and maintains essential biogeochemical cycles due to the action of microorganisms. Although the soil is inhabited by populations including macrofauna, mesofauna, microfauna, and microflora, it is estimated that 80-90% of the processes occurring in soil, including nutrient recycling, are mediated by microorganisms. Statistical data indicate that 1 gram of soil can contain about 10 billion microorganisms. However, 95-99% of these microorganisms can only be identified by molecular techniques, independent of cultivation. Currently, the reduction in costs of generating molecular data and advances in high-throughput technologies have demanded the training of new professionals in agriculture, with knowledge in bioinformatics, capable of understanding and analyzing the growing amount of molecular data. In this perspective, the present project aims to develop and apply analytical and computational methods in genomics, proteomics, metagenomics, phylogeny, and soil microbial ecology studies. Research Line: Applied Computational Modeling.
Researchers:
Luciano Jose Senger (Project Leader)
Description:
Parallel and distributed computing aims to solve computational problems through the use of interconnected computers, which work together to resolve a single issue. This approach allows for cost reduction and improved performance of computational processes. The main goal of this thematic project is to study parallel and distributed processing techniques for the development of software that enables applications to run on different platforms. Among the existing distributed platforms for computing are multi-core processors (dual/quad cores), clusters of personal computers, massively parallel processing machines, computational grids, and peer-to-peer (P2P) networks. Other objectives of the project include theoretical study of performance evaluation models and application scheduling, using computational simulation techniques and practical experimentation. With the implementation of the master's program in applied computing at UEPG, this thematic project will also focus efforts on developing parallel applications for agricultural problems, such as distributed data mining and optimization through parallel algorithms and heuristics. Currently, efforts are directed towards applying parallel processing techniques to non-specific problems, through the definition of generic software libraries. Thus, there is extensive knowledge in research and software development that can be directly applied to solving problems in agriculture.
This thematic project includes two main projects, supported respectively by CNPq and Fundação Araucária. The first project, titled “Simulated Annealing Metaheuristics for Knowledge-Based Process Scheduling in Distributed Computing Systems,” proposes a study on the use of the simulated annealing metaheuristic (SA) for knowledge-based scheduling in distributed computing systems. This algorithm uses application execution characteristics obtained through feature extraction models to guide metaheuristic decisions. Effective scheduling is extremely important for improving performance in parallel systems.
The second project, titled “A Framework for Parallel and Distributed Computing in P2P Networks,” was conceived based on the limitations identified in developing distributed applications over P2P networks due to the complexity of designing and implementing these applications on such networks. This project proposes defining and implementing a framework to facilitate the development and execution of applications in P2P networks. The framework will consist of a set of programming libraries to be used in implementing distributed applications and management, monitoring, and scheduling software for parallel and distributed processes created using this library.
This thematic project involves researchers from the Department of Computer Science at UEPG, the Institute of Mathematical and Computational Sciences at USP (ICMC/USP), and St. Francis Xavier University, Canada. Research results in performance evaluation and application scheduling have been published in field journals and include the following awards: Outstanding Paper Award at IEEE International Conference on Computational Science and Engineering (2008); IEEE Outstanding Paper Award: 21st IEEE International Conference on Advanced Information Networking and Applications (AINA-07) (2007); IEEE Computer Society 2006 Best Paper Award: IEEE International Conference on Advanced Information Network and Applications (AINA 2006), IEEE Computer Society (2006).
Research Line: Applied Computational Modeling
Researchers:
Alaine Margarete Guimarães (Project Leader)
Arion de Campos Junior
Eduardo Fávero Caires
José Carlos Ferreira da Rocha
Description:
Data mining (DM) research, which aims to develop and use automatic techniques to explore large amounts of data to discover new patterns and relationships, evolves in different directions. One of the challenges in DM is analyzing data with characteristics that are considered difficult for knowledge discovery, including: imbalance, real numerical attributes, numerous attributes, and time series. These characteristics are present in a range of agronomic databases, including precision agriculture databases and agroclimatic data. Precision agriculture, due to the large volume of physical and chemical soil data generated, combined with their georeferenced profile, demands the development of intelligent analysis techniques and pattern identification applicable to such data. The use of agroclimatic data is present in various agricultural segments. However, these data are often large in volume, sometimes with errors and observation failures, requiring careful analysis and usage.Moreover, associating these data with information from different crops can establish edaphoclimatic relationships that contribute to predicting plant diseases. The Infoagro laboratory research group has been developing research to find relationships between agroclimatic variable data associated with soybean and wheat crops, aiming to predict the likelihood of aggressor agents in the crops. In this context, data mining can help identify non-obvious behavior patterns in crops due to environmental conditions.Thus, agriculture presents complex databases that constitute challenges for the field of Data Mining, while also requiring intelligent data analysis methods to provide advances and strategic decision-making.The objective of this project is to study, develop, and apply Data Mining (DM) methods for knowledge discovery in agricultural databases, considering those involving physical and chemical soil data, agroclimatic data, imbalanced data, temporal data, and georeferenced data.The Applied Computing Master's Program has several faculty members conducting research related to data mining and agricultural data analysis, which provides strong integration for this thematic project, enabling research on: Determining objective and subjective functions for agricultural data mining; Using Data Mining techniques to determine nutrient criteria in direct planting systems; Evolutionary Computation in Agricultural Data Mining; Using Neural Networks, Fuzzy Theories, and Bayesian Networks in agricultural data mining; Hybrid Systems; Distributed Systems for data mining; Multi-objective Optimization by Particle Swarm and Data Mining; Temporal Data Mining; Georeferenced Data Mining; Visualization of rules based on georeferenced data. Research Line: Computing, Automation, and Data Management in Agriculture.
Researchers:
Alaine Margarete Guimarães (Project Leader)
Alceu de Souza Britto Junior
José Carlos Ferreira da RochaRosane Falate
Description:
In agriculture, the proper choice of machinery and equipment is essential to achieve maximum production yield. As one of the first steps toward satisfactory production, the sowing process must be well planned, considering the distribution of seeds, including precision row planting, where seeds are metered, with a uniform spacing allowed, provided the average population density is within the optimal range for that variety.A factor that directly affects seed uniformity is the type of metering mechanism, with perforated disk types being the most commonly used. To choose the appropriate metering disk, among other methods, analyses are performed that simulate seed distribution in the soil, which involves measuring the spacing between subsequent seeds and checking their arrangements. This spacing is classified as acceptable, double, and missed, according to a predefined minimum and maximum interval. Based on these criteria, various perforated disks are tested for the desired seed, and the one with the best performance in distribution should be considered. However, the currently adopted analysis procedure, which is based on visual observation of spacings, has proven to be rudimentary and manual in most cases, leading to overload and potential errors in the process. Some automation attempts have been developed but do not show efficient and satisfactory behavior for effective adoption.The coordinator of this project, together with the INFOAGRO/UEPG Laboratory team and the company PLÂNTULA-SOCIDISCO, developed a prototype system to automate the seed distribution analysis process in sowing procedures using a video camera and Digital Image Processing (DIP) techniques, aiming to implement a simple, easy-to-operate, and low-cost system. Thus, a method was designed for analyzing seed spacing in an environment consisting of a personal computer with an image capture card, a low-cost camcorder, and a simplified lighting system, coupled with a seed distribution simulation device equipped with a conveyor and a metering mechanism.Experiments indicated that sowing analysis using a video camera and DIP techniques could be feasible, with two issues found in the literature, the identification of doubles in distribution and the recognition of very small seeds, being resolved with the image processing techniques used. It is necessary to improve the accuracy in identifying smaller and adjacent seeds.A second work developed by the team consists of a seed distribution evaluation system based on sensors that indicate the presence or absence of seeds and an information processing system that counts the elapsed time between seed detection events on the conveyor. The longitudinal distribution was determined using a normal reflection sensor and the RCX detection and actuation module from the LEGO MINDSTORMS 9793 kit, also available at LAR. The result was satisfactory, and now the development of a dedicated integrated circuit for this task should be studied.The objective of this research and technological development project is, based on previously conducted studies, to develop an embedded intelligent system for automating the longitudinal seed distribution analysis in a pre-planting environment, aiming to obtain more appropriate solutions in terms of cost and performance. This project involves studies in agricultural automation, computational intelligence, and agricultural systems modeling, involving faculty researchers, undergraduate students, and graduate students from the Agronomy, Computer Science, and Computer Engineering programs. Research Line: Computing, Automation, and Data Management in Agriculture.