Loading...
22 results
Search Results
Now showing 1 - 10 of 22
- Optimizing Olive Disease Classification Through Hybrid Machine Learning and Deep Learning TechniquesPublication . Mendes, João; Moso, Juliet; Berger, Guido; Lima, José; Costa, Lino; Guessoum, Zahia; Pereira, Ana I.Olive trees play a crucial role in the global agricultural landscape, serving as a primary source of olive oil production. However, olive trees are susceptible to several diseases, which can significantly impact yield and quality. This study addresses the challenge of improving the diagnosis of diseases in olive trees, specifically focusing on aculus olearius and Olive Peacock Spot diseases. Using a novel hybrid approach that combines deep learning and machine learning methodologies, the authors aimed to optimize disease classification accuracy by analyzing images of olive leaves. The presented methodology integrates Local Binary Patterns (LBP) and an adapted ResNet50 model for feature extraction, followed by classification through optimized machine learning models, including Stochastic Gradient Descent (SGD), Support Vector Machine (SVM), and Random Forest (RF). The results demonstrated that the hybrid model achieved a groundbreaking accuracy of 99.11%, outperforming existing models. This advancement underscores the potential of integrated technological approaches in agricultural disease management and sets a new benchmark for the early and accurate detection of foliar diseases.
- Deep learning method to identify fire ignitionsPublication . Mendes, João; Brito, Thadeu; Pereira, Ana I.; Lima, JoséThe SAFe project aims to create and implement a set of innovative operations that minimize the time of forest ignitions identification contributing to the development of the Trás-os-Montes region. Thus, it is intended to locate a set of sensors in the forest, data information will be collected, and the artificial intelligence algorithm Deep Learning will be applied to achieve the intended end. Numerical results demonstrated the approach reliability.
- Deep learning networks for olive cultivar identification: a comprehensive analysis of convolutional neural networksPublication . Mendes, João; Lima, José; Costa, Lino; Rodrigues, Nuno; Pereira, Ana I.Deep learning networks, more specifically convolutional neural networks, have shown a notable distinction when it comes to computer vision problems. Their versatility spans various domains, where they are applied for tasks such as classification and regression, contingent primarily on the availability of a representative dataset. This work explores the feasibility of employing this approach in the domain of agriculture, particularly within the context of olive growing. The objective is to enhance and facilitate cultivar identification techniques by using images of olive tree leaves. To achieve this, a comparative analysis involving ten distinct convolutional networks (VGG16, VGG19, ResNet50, ResNet152-V2, Inception V3, Inception ResNetV2, XCeption, MobileNet, MobileNetV2, EfficientNetB7) was conducted, all initiated with transfer learning as a common starting point. Also, the impact of adjusting network hyperparameters and structural elements was explored. For the training and evaluation of the networks, a dedicated dataset was created and made available, consisting of approximately 4200 images from the four most representative categories of the region. The findings of this study provide compelling evidence that the majority of the examined methods offer a robust foundation for cultivar identification, ensuring a high level of accuracy. Notably, the first nine methods consistently attain accuracy rates surpassing 95%, with the top three methods achieving an impressive 98% accuracy (ResNet50, EfficientNetB7). In practical terms, out of approximately 2016 images, 1976 were accurately classified. These results signify a substantial advancement in olive cultivar identification through computer vision techniques.
- A machine learning approach to robot localization using fiducial markers in robotatfactory 4.0 competitionPublication . Klein, Luan C.; Braun, João; Mendes, João; Pinto, Vítor H.; Martins, Felipe N.; Oliveira, Andre Schneider; Oliveira, Andre Schneider; Wörtche, Heinrich; Costa, Paulo Gomes da; Lima, JoséLocalization is a crucial skill in mobile robotics because the robot needs to make reasonable navigation decisions to complete its mission. Many approaches exist to implement localization, but artificial intelligence can be an interesting alternative to traditional localization techniques based on model calculations. This work proposes a machine learning approach to solve the localization problem in the RobotAtFactory 4.0 competition. The idea is to obtain the relative pose of an onboard camera with respect to fiducial markers (ArUcos) and then estimate the robot pose with machine learning. The approaches were validated in a simulation. Several algorithms were tested, and the best results were obtained by using Random Forest Regressor, with an error on the millimeter scale. The proposed solution presents results as high as the analytical approach for solving the localization problem in the RobotAtFactory 4.0 scenario, with the advantage of not requiring explicit knowledge of the exact positions of the fiducial markers, as in the analytical approach.
- A saúde mental na população das instituições associadas da CNISPublication . Correia, Helder; Fernandes, Adília; Pereira, Ana I.; Mata, Maria Augusta; Magalhães, Carlos Pires; Rodrigues, Clementina; Mendes, JoãoA presente publicação decorre de um estudo, no qual se procurou identificar a perceção quanto aos efeitos das contingências inerentes à Pandemia SARS-CoV2 na saúde mental de utentes e colaboradores, das Instituições Associadas da CNIS.
- Data acquisition filtering focused on optimizing transmission in a LoRaWAN network applied to the WSN forest monitoring systemPublication . Brito, Thadeu; Azevedo, Beatriz Flamia; Mendes, João; Zorawski, Matheus; Fernandes, Florbela P.; Pereira, Ana I.; Rufino, José; Lima, José; Costa, Paulo Gomes daDeveloping innovative systems and operations to monitor forests and send alerts in dangerous situations, such as fires, has become, over the years, a necessary task to protect forests. In this work, a Wireless Sensor Network (WSN) is employed for forest data acquisition to identify abrupt anomalies when a fire ignition starts. Even though a low-power LoRaWAN network is used, each module still needs to save power as much as possible to avoid periodic maintenance since a current consumption peak happens while sending messages. Moreover, considering the LoRaWAN characteristics, each module should use the bandwidth only when essential. Therefore, four algorithms were tested and calibrated along real and monitored events of a wildfire. The first algorithm is based on the Exponential Smoothing method, Moving Averages techniques are used to define the other two algorithms, and the fourth uses the Least Mean Square. When properly combined, the algorithms can perform a pre-filtering data acquisition before each module uses the LoRaWAN network and, consequently, save energy if there is no necessity to send data. After the validations, using Wildfire Simulation Events (WSE), the developed filter achieves an accuracy rate of 0.73 with 0.5 possible false alerts. These rates do not represent a final warning to firefighters, and a possible improvement can be achieved through cloud-based server algorithms. By comparing the current consumption before and after the proposed implementation, the modules can save almost 53% of their batteries when is no demand to send data. At the same time, the modules can maintain the server informed with a minimum interval of 15 min and recognize abrupt changes in 60 s when fire ignition appears.
- Deep Learning na identificação de incêndios florestaisPublication . Mendes, João; Pereira, Ana I.O problema dos incêndios florestais, é uma constante mundial visto ainda não se ter encontrado uma solução suficientemente eficaz para o seu combate e irradicação. Um dos grandes problemas deste tema no nosso país é o facto da floresta portuguesa possuir uma vegetação muito densa, o que limita o uso de sensores de imagem que segundo alguns autores são a melhor opção para a deteção de ignições florestais. Uma vez que esta opção de vigilância é excluída e considerando que os primeiros 20 minutos são essenciais para minimizar os danos causados pelo incêndio, deverão ser estudadas e implementadas técnicas mais inovadoras de maneira a combater este flagelo. Para isso, o projeto SAFe apresenta como solução um conjunto de operações inovadoras que minimizam o tempo de identificação de ignições florestais contribuído assim, em última instância, para o desenvolvimento da região de Trás-os-Montes. Assim, pretende-se implementar um conjunto de sensores na floresta, onde serão recolhidos os dados referentes as variáveis mais relevantes para a identificação de ignições florestais, uma vez recolhidos esses dados serão analisados através de um sistema de inteligência artificial e por fim serão enviados os alertas as entidades competentes. O protótipo da caixa de recolha de dados é composto por vários tipos de sensores, nomeadamente sensores de deteção de chama, sensores de temperatura, humidade do solo e do ar e sensores de radiação ultravioleta. O processo de transmissão deste tipo de sensores é assegurado através da comunicação LoRa, uma vez que existe a necessidade de transmitir grandes quantidades de dados (os sensores recolhem dados constantemente em intervalos de dois minutos). Este sistema permitirá obter um controlo instantâneo, facilitando assim a tarefa de processamento e posterior alerta dentro da janela de tempo necessária para minimizar os danos causados pelo incêndio. O uso da inteligência artificial, mais propriamente do algoritmo Deep Learning é uma opção neste caso devido a quantidade de dados obtida no dia a dia, produzindo assim um sistema que se irá tornar mais inteligente a cada hora que esteja a funcionar, devido ao treino constante, o que resultara num software mais fiável dia após dia.
- Pest Management in Olive Cultivation Through Computer Vision: A Comparative Study of Detection Methods for Yellow Sticky TrapsPublication . Mendes, João; Berger, Guido; Lima, José; Costa, Lino; Pereira, Ana I.This study compares two computer vision methods to detect yellow sticky traps using unmanned autonomous vehicles in olive tree cultivation. The traps aim to combat and monitor the density of the Bactrocera oleae, an important pest that damages olive fruit, leading to substantial economic losses annually. The evaluation encompassed two distinct methods: firstly, an algorithm employing conventional segmentation techniques like thresholding and contour localization, and secondly, a contemporary artificial intelligence approach utilizing YOLOv8, a state-of-the-art technology. A specific dataset was created to train and adjust the two algorithms. At the end of the study, both were able to locate the trap precisely. The segmentation algorithm demonstrated superior performance at proximal distances (50 cm), outperforming the outcomes achieved by YOLOv8. In contrast, YOLOv8 exhibited sustained precision, irrespective of the distance under examination. These findings affirm the versatility of both algorithms, highlighting their adaptability to various contexts based on distinct application demands. Consideration of trade-offs between accuracy and processing speed is essential in determining the most appropriate algorithm for a given application.
- Using machine learning approaches to localization in an embedded system on RobotAtFactory 4.0 competition: a case studyPublication . Klein, Luan C.; Braun, João; Martins, Felipe N.; Wörtche, Heinrich; Oliveira, Andre Schneider; Mendes, João; Pinto, Vítor H.; Costa, Paulo Gomes daThe use of machine learning in embedded systems is an interesting topic, especially with the growth in popularity of the Internet of Things (IoT). The capacity of a system, such as a robot, to self-localize, is a fundamental skill for its navigation and decision-making processes. This work focuses on the feasibility of using machine learning in a Raspberry Pi 4 Model B, solving the localization problem using images and fiducial markers (ArUco markers) in the context of the RobotAtFactory 4.0 competition. The approaches were validated using a realistically simulated scenario. Three algorithms were tested, and all were shown to be a good solution for a limited amount of data. Results also show that when the amount of data grows, only Multi-Layer Perception (MLP) is feasible for the embedded application due to the required training time and the resulting size of the model.
- Machine learning to identify olive-tree cultivarsPublication . Mendes, João; Lima, José; Costa, Lino; Rodrigues, Nuno; Brandão, Diego; Leitão, Paulo; Pereira, Ana I.The identification of olive-tree cultivars is a lengthy and expensive process, therefore, the proposed work presents a new strategy for identifying different cultivars of olive trees using their leaf and machine learning algorithms. In this initial case, four autochthonous cultivars of the Trás-os-Montes region in Portugal are identified (Cobrançosa, Madural, Negrinha e Verdeal). With the use of this type of algorithm, it is expected to replace the previous techniques, saving time and resources for farmers. Three different machine learning algorithms (Decision Tree, SVM, Random Forest) were also compared and the results show an overall accuracy rate of the best algorithm (Random Forest) of approximately 93%.
- «
- 1 (current)
- 2
- 3
- »