Masters Theses and Dissertations

Permanent URI for this collection${dspace.ui.url}/handle/20.500.12479/33

Browse

Now showing 1 - 20 of 172

Development of an IoT-based smart irrigation system for efficient water management in Uasin Gishu County
(NM-AIST, 2024-08) Bundotich, Winny
Agriculture is the backbone of Kenya’s economy, with Uasin Gishu county being the country’s breadbasket. Food scarcity has recently increased due to climate change, population growth, and decreased land available for farming. Several measures have been implemented to mitigate food scarcity, including encouraging irrigation farming to ensure whole-year food production. However, irrigation practice faces a water scarcity challenge. Efforts have been put in place to improve water use efficiency. These measures include scheduled irrigation system technology. Most of these scheduled systems target greenhouse irrigation and leave out open-field irrigation farmers where rainfall is a factor in reducing water wastage whenever there is rainfall. This technology does not provide a precision of plant water needs and poses a risk of under or over irrigation. Therefore, to overcome these challenges, this project developed an IoT-based system with sensors to monitor critical soil parameter measurements continuously. The main objective of this project was to develop an IoT-based smart irrigation system for efficient water management. An ESP32 microcontroller board is used to process information collected from the sensors. Openweather API is implemented on the Thingsboard cloud platform to fetch rainfall prediction information. While providing remote valve control, the system uses soil moisture level and rainfall prediction parameters to automatically control the irrigation valves. The system also offers rich farm information visualization through the Thingsboard cloud platform dashboard, which can be accessed remotely through the Thingsboard live mobile application, where a user can control the irrigation valves remotely. Mixed methods which involved questionnaires and focus group discussions was used to collect data from sixteen respondents. Purposve sampling was also used to identify the respondents during data collection. To develop the system,agile software development methodology, specifically extreme programming (XP) was implemented. To validate the system’s functionalities, the system was demonstrated to twenty seven people and thereafter were allowed to interact with the system. A questionnaire was implemented to get feedback from the respondents. Generally, respondents agreed that the system satisfactorily met their needs. The developed system contributes to the value chain by providing precise water input. The system can be advanced to include other essential features needed to monitor and evaluate irrigated farms within Uasin Gishu county and any other region.
IoT-based automated surface-drip irrigation monitoring system prototype: a case of World Vegetable Center in Arusha, Tanzania
(NM-AIST, 2024-08) Wangere, Joseph
Irrigation is pivotal for supplementing water, amidst climate change globally and surface-drip irrigation provides an efficient plant-watering method with minimal mineral depletion. In Tanzania, Africa, manual surface-drip irrigation persists despite attempts at SMS-based automation, leading to time-consuming routine irrigation tasks. Existing studies focus on XBee modules in their transparent operation mode, limiting scalability due to lack of information such as source addresses as well as lack of support for multiple sensor nodes. This project, carried out in Arusha, Tanzania, employed a wireless sensor network using XBee modules configured in API mode for soil-data collection and transmission, utilizing Node-RED for remote actuator control and soil data visualization. By employing JavaScript functions, data bytes were extracted, including transmitter addresses, from each sensor node, enhancing scalability. Sensors measured soil moisture, temperature, humidity, air temperature, and ultraviolet radiation to determine irrigation requirements. The outcome, an IoT-based Automated Surface-drip Irrigation Monitoring System Prototype was tested against user and system requirements. Remarkably, the system reduced irrigation control time by 96%, compared to manual control, by automatically triggering on the pump and valves when minimum soil moisture threshold is reached and turning them off once the maximum moisture threshold is reached. It also supported over-the-air firmware updates, to reduce maintenance costs, enabling the integration of technologies like digital twins. The system can integrate with existing manual irrigation systems, offering a cost-effective automation solution. Future improvements could involve incorporating machine learning for crop water requirement forecasting, expanding the sensor node network and extensive testing under different farm environments.
Industrial-based GSM water leakage detection, monitoring, and controlling system: a case of North Rift Valley Water Agency in Kenya
(NM-AIST, 2024-06) Kipketer, Dickson
The current system for detecting and monitoring water leaks in Kenyan industries is manual and costly. Despite emerging technological trends, many industries lack automated systems to detect, monitor, and control water leakage due to high maintenance and installation costs. This study aimed to develop an automatic, remote, real-time detection, monitoring, and control of water leaks in North Rift Valley Water Works Agency. The system is made up of two nodes, one at the source and another at the destination or tap. The two nodes are made up of an ESP32 Microcontroller, which is used to control all the connected components. The ESP32 Microcontroller was efficient due to its ability to provide WI-FI. Aside from the solenoid valve, which was used to turn the water flow on or off in the event of leaks, the system also includes the FY-201 water flow sensor, which was used to gauge the amount of water flowing through the pipe. Water leakage is detected when the water passing through the two sensors differs slightly, indicating that a water leakage has just occurred. Thing Board, an IoT-based platform used to monitor and visualize data from various connected devices, was used for real-time monitoring, visualization, and control. The system administrator could log into and manage the system by remotely turning the water leaks on and off from their phone. The database used was MySQL DB, and a system was created using C programming language, the Arduino Integrated Development Environment, HTML, and Cascading Style Sheet (CSS). The developed system was tested with different water service providers, including Eldoret Water and Sanitation Company, and the results show that the system responds positively to water leakage parameters. The developed system could monitor water leakages in real-time with the two nodes interacting by sharing information via the server and communicating via WI-FI enabled by the ESP 32 Microcontroller. With new technological trends, such as the Internet of Things, these microcontrollers will facilitate faster adoption in industrial projects. The benefits include that a system administrator can control the system remotely without physically switching the solenoid valve on or off. It is also lower in cost compared to other tariffs. The system can be installed in hazardous areas, such as valleys and mountains, making it attractive. Future research suggests more investigation into identifying leaks at particular points along water pipes by integrating with the Global Positioning System (GPS) to determine their precise location.
Web-based freight forwarding system for logistics management: case study of Trueline Africa Limited in Kampala, Uganda
(NM-AIST, 2024-08) Mary, Siama
Freight forwarding sector is moving from manual handling of freights to electronic freight because of the ever-changing needs of technologies. The study aimed to develop a freight forwarding web-based information system to manage logistics by automating manual processes that enable electronic data exchange, accelerate processes, improve communication, facilitate easy access to and retrieval of documents, enhance customer service, and drive efficiency and profitability. In order to collect the data, the author employed both qualitative and quantitative techniques, as well as the Extreme Programming (XP) system development methodology. The developed system was tested and assessed using unit testing, integrated testing, and system testing. By creating a survey questionnaire and distributing it to the system's end users, the system was validated using the TAM with Trust. The verdicts showed that perceived usefulness, perceived ease of use, and system acceptance are all influenced by trust, which is believed to be a key component. Furthermore, it has been found that system usage and perceived utility are greatly enhanced by perceived ease of use. The creation of this system has reduced errors, work overload, operating expenses, and the likelihood of on-time item delivery. The author advises using the Global Positioning System (GPS) for cargo tracking and visibility in additional investigations, as well as integrating it with customs systems for compliance checks and guaranteeing adherence to legal standards to lower the danger of fines and delays.
Time series and ensemble models for forecasting Tanzanian banana crop yield under various effects of Climate change
(NM-AIST, 2024-07) Patrick, Sabas
Amid escalating global worries about climate change’s impact on agriculture, this study thor oughly explores how climate shifts might affect Tanzania’s vital bananas. The study employed a multiple regression model to analyze the correlation between bananas and key climate vari ables in Tanzania, the results showed gradual decrease in bananas. Additionally, the study utilized two powerful global sensitivity analysis methods, Sobol’ Sensitivity Indices and Re sponse Surface Methodology, to comprehensively explore the sensitivity of bananas to climate variables. So, these methods showed that minimum temperature, precipitation and soil moisture have the most impact on bananas and affect the crop’s production variability. Furthermore, un certainty quantification was performed using Monte Carlo simulation, estimating uncertainties in regression model parameters to enhance the reliability of findings, this indicated substantial variability in the predictions. Conversely, the study configured time series models such as Sea sonal ARIMAwithExogenousVariables (SARIMAX),State Space (SS), and Long Short-Term Memory (LSTM) to forecast bananas in Tanzania under the effects of climate change. Hence, the study builds predictive frameworks capturing temporal variations and offering glimpses of future trends. Leveraging historical bananas data and relevant climate variables, an ensemble model was formulated using a weighted average approach, revealing a future decrease in ba nanas. This study combines data analysis and advanced models to explore how climate change affects bananas. Its insights reach beyond farming, impacting stakeholders, policymakers, and farmers alike. By understanding sensitivities, vulnerabilities, and future trends, this research informs decisions for sustainable banana production, enhances food security, and encourages adaptable strategies amidst changing climates.
Modelling the dynamics of Spodoptera frugiperda infestation in maize production with control strategies
(NM-AIST, 2024-08) Reuben, Yusuph
The ongoing demand for maize is driven by its nutritional value, ability to meet the food re quirements of a increasing world population, its impact on ensuring a stable food supply, and the growing international investments in ethanol as a renewable fuel source. Nevertheless, the invasion and extensive spread of the Spodoptera frugiperda pest cause substantial losses in maize yields, resulting in a diminished standard of living and an economic downturn for those involved in maize production. This study formulates the systems of differential equations to simulate the dynamics of a model encompassing both Spodoptera frugiperda and maize biomass. The model considers the presence of predators and the application of best farming practices. The model demonstrates six equilibrium points, and they are locally asymptotically stable provided that the essential conditions are satisfied. Global stability for models’ equi libria are established by Lyapunov functions. Latin Hypercube Sampling (LHS) and Partial Rank Correlation Coefficient (PRCC) multivariate analysis were utilized to pinpoint the sen sitive parameters that influence the pest. The predators attack rate and best farming practices was the most sensitive parameter, when increases maize biomass increases while pest popu lation significantly dencreases. Numerical simulations indicate that, during the initial phases, combining natural enemies with best farming practices emerges as a successful intervention by directly decreasing the pest population and fostering sustainable pest control. The optimal control strategy which combines farming education campaign and predator selection awere ness on eggs and larval reduced pest infestation in early stages. Thus, to control the pest, we recommend that more efforts be directed to reduce eggs and larval populations and improving farming methods through education campaign.
An automated core rolling machine for making the metering unit’s current transformer
(NM-AIST, 2024-07) Magomere, Rebecca
The advent of automation has revolutionized various industries by simplifying complex processes, enhancing efficiency, and reducing production time. However, many developing countries still rely on manual labour, resulting in lower production rates and reduced competitiveness in the global market. This study focuses on the challenges faced by Tanzania's electrical equipment company (TANELEC) in the production of metering units, specifically the laborious and hazardous task of rolling the core steel for the units' current transformer (CT). Manual processes often lead to dimensional inaccuracies and gaps in the rolled core, causing a significant number of tested metering units to be deemed unqualified. Consequently, the industry struggles to meet customer demands effectively. The research findings emphasize the urgent need for automating the core rolling process for metering units' CT, which became the primary objective of this study. Employing a qualitative research approach and utilizing the agile methodology, specifically the Extreme Programming agile method, the study aimed to gain a comprehensive understanding of the current manual process. The outcome of the study is a machine capable of intelligently accepting and rolling the required ferromagnetic iron material, with users able to input desired dimensions and quantities of cores. This development has significantly increased TANELEC's production capacity, enabling the manufacturing of up to sixty cores per hour and alleviating the production burden. As a result, the company can now meet customers' demands promptly.
Mathematical models for vehicular carbon dioxide emission
(NM-AIST, 2024-08) Donald, Pita
The increasing demand for transportation due to a growing global population has led to more vehicles on the road and increased use of fossil fuels, resulting in higher atmospheric carbon dioxide (CO2) levels and contributing to global warming. Thus, adopting sustainable trans portation practices is crucial for achieving climate change goals, specifically the reduction of greenhouse gas emissions to mitigate global warming. This study presents a nonlinear mathe matical model to analyze the dynamics and control of atmospheric CO2 concentration in rela tion to vehicle emissions. The model is qualitatively analyzed to understand long-term system behavior. Model parameters are calibrated using real-world data on world population, eco nomic activities, atmospheric CO2, forest biomass, and vehicle numbers. Results describes the dependence between vehicle CO2 emissions and atmospheric CO2 levels and impact human population decline. Numerical simulations validate analytical findings, and global sensitivity analysis explores the influence of various parameters on CO2 dynamics. An optimal control problem is formulated and solved by using Pontryagin’s principle, establishing optimality con ditions. Solving the problem reveals that reducing vehicle emissions, implementing reforesta tion efforts, adopting green economy practices, and curbing fossil-fueled vehicle production can cut atmosphericCO2 levels by 2.866%. Consequently, addressing climate change linked to increased atmospheric CO2 concentration is achievable through these measures.
Modeling of diamondback moth infestations in cabbages with control strategies
(NM-AIST, 2024-08) Paul, Daniel
The Diamondback moth (DBM), scientifically called Plutella xylostella, a highly destructive and rapidly spreading agricultural pest originally from Europe. This particular pest presents a notable risk to the overall global food security, with estimates suggesting that periodic out breaks of DBM lead to annual crop losses of up to $US 4−5 billion worldwide. Given the potential for such substantial losses, it is crucial to employ various methods and techniques to understand the factors affecting the interaction between DBM and cabbages, which, in turn, impact cabbage biomass. In this research, two mathematical models were developed to assess the effects of pesticides, inter-cropping methods, and an Integrated Pest Management strategy on DBMinfestations in cabbage biomass. The first model focused on the impact of pesticides through nonlinear Ordinary Differential Equations (ODEs) to analyze the interaction between DBMandcabbage biomass. The model’s results indicated that the use of pesticides effectively eliminate DBM in a cabbage farm. The second nonlinear ODEs model assessed the impact of providing pest control education to farmers and deploying predators in cabbage farms. We found that provision of pest control education alone could reduce the negative impact of the DBM on final cabbage biomass by 10%. However, when both tactics were implemented to gether, the pest population was reduced by 40%–80%. In both proposed models, the MATLAB was employed for simulations to affirm the validity of the analytical results. In brief, the find ings of this research highlighted the importance of mathematical models as useful instruments for comprehending the management of DBM in cabbage crops.
Biometric banking system for effective paperless cash withdrawals at CRDB Bank Burundi
(NM-AIST, 2024-08) Niyongabire, Pacifique
In the ever-changing banking landscape, attempts have been made to improve customer experiences and expedite operations. However, there remains a continuing difficulty at bank teller counters, where transactions are still done on paper, resulting in operational inefficiency and customer dissatisfaction. Despite its effort to provide excellent financial services, CRDB Bank Burundi confronts a similar difficulty. To remedy this, our project introduces a cutting edge biometric banking system for secure and paperless withdrawals at the bank teller desk. The development integrates modern biometric technology, such as fingerprint scanning and RFID card identification for customers with hand disabilities or contactless alternative. It is guided by a detailed investigation of existing banking inefficiencies. The system promotes client satisfaction by allowing personal interaction with bank tellers and integrating clear notifications via SMS, WhatsApp, and email after each transaction. Furthermore, the technology links fingerprints or cards to current bank accounts, allowing for faster withdrawals and a more secure banking environment. Extensive testing proves the system's dependability, efficiency, streamlined operations, and other features contributing to its effectiveness, paving the way for deployment. This initiative represents a paradigm leap in banking, offering clients a smooth and secure transaction experience.
Mathematical modeling and extraction of parameter of photovoltaic module based on modified Newton-Raphson method
(NM-AIST, 2024-08) Mlazi, Nsulwa
Photovoltaic (PV) generators are represented by varoius types of electrical equivalent circuits. Each of which describes the output current-voltage relationship under particular operating con ditions. Ideal model, single and double diode models are examples to representation of photo voltaic cell/module. In order to assess the performance of PV generators, one needs to extract essential parameters of PV module. This dissertation introduces a numerical approach for estimating four crucial physical parameters within a single-diode circuit model based on the manufacturer’s datasheet. The methodology involves establishing a system of four non-linear equations derived from three pivotal points in PV characteristics. Through suggested iterative approach, the photocurrent, saturation current, ideality factor, and series resistance are deter mined utilizing the proposed method. Validation of the suggested technique is conducted using RTCFrance solar cell, Chloride CHL285P, and Photowatt PWP210 modules. The obtained re sults are compared with in-field outdoor measurements, demonstrating a commendable agree ment with the experimental data. Furthermore, the selected model is subjected to simulation in the MATLAB environment to evaluate its response to external physical weather conditions, specifically temperature and solar irradiance. Notably, the proposed method exhibits a faster convergence compared to the widely utilized Newton method, emphasizing its efficiency. The significance of modeling PV cells/modules is underscored as it plays a crucial role in predicting the performance of photovoltaic generators under varying operating conditions. This numer ical method contributes to the field by offering a quicker convergence compared to existing techniques, thereby enhancing its practical utility.
A predictive analytics-driven battery management system for sustainable e-mobility in East Africa
(NM-AIST, 2024-08) Ochiel, Michael
In pursuit of the United Nations Sustainable Development Goals (UN SDGs) seven and thirteen, East African countries are swiftly transitioning to electric mobility solutions for clean transportation and climate action. However, this transition presents a challenge in repurposing and maintenance of used electric vehicle (EV) batteries due to limited specialized knowledge and equipment in the region. Despite the growing popularity of electric vehicles, a significant gap exists in understanding viable battery components for second life applications in East Africa. This study addresses this gap by designing a predictive analytics-driven battery management system tailored to the region's needs. The developed system integrates hardware and software, employing a data-driven approach to analyze sensor data for decision support and enable remote monitoring of repurposed batteries. Compared to existing works, this research emphasizes the use of predictive algorithms to monitor battery health in second life applications and provision for remote monitoring. This innovative approach significantly advances the understanding and implementation of battery repurposing in East Africa. By offering a sustainable solution for e-mobility, this study promotes a cleaner and greener future while reducing energy costs for organizations and domestic users.
A hyperspectral-based system for identification of common bean genotypes resistant to foliar diseases in Tanzania
(NM-AIST, 2024-06) Irakiza, Josiane
Common bean is one of many legumes (family Fabaceae) widely cultivated for their edible seeds, seedpods, and leaves. Despite its benefits and life dependency especially in sub-Saharan Africa, foliar diseases are causing a loss of 20% to 80% of common bean production, and the development of improved common bean seeds resilient to those foliar diseases is still an issue where among the major problem that the bean breeders are facing is manual phenotyping; a slow field process and prone to errors as it depends on the eyes of the viewer. According to the literature, imaging technologies have been introduced to help in different processes for crops and disease management. However, there is a lack of automated mechanisms for phenotyping processes to help breeders in trait data collection, disease scare classification, and analysis of all data collected to identify resilient genotypes digitally. Among existing solutions, there is still also a gap in plant health monitoring during all its growing stages needed by breeders. Therefore, this study developed a unique hyperspectral data-based approach for identifying bean genotypes resistant to foliar diseases and plant health monitoring. Using the Random Forest classifier this study proved the genotype classification in three main breeding categories; Resistant, Medium, and Susceptible. The experiment was conducted in four Regions of Tanzania and three classifiers “Random Forest, XgBoost, and four layers Neuro Network algorithms” were trained and tested with results of 0.96, 0.95, and 0.92 respectively. The model was deployed on the cloud server where it is linked to a web application for easy classification and data analysis. Applying different vegetation indexes including the Chlorophyll Index, Photochemical Reflectance Index, Water Band Index, Modified Chlorophyll Absorption in Reflectance Index, Nitrogen Reflectance Index, Structure Insensitive Pigment Index, and Simple Ratio efficiently proved to be used for plant health insight before disease symptoms are seen. This saves breeders time, reduces errors, and helps them with digital phenotypic data, faster analysis, and easy storage for future references.
Development of mobile-based recommender system for smallholder dairy farmers to increase their production in Tanzania
(NM-AIST, 2024-03) Malamsha, Glory
Dairy farming is a branch of agriculture which is devoted to the production of milk and the processing of dairy products. Small holder farmers are individuals producing small amounts of products on small land holdings of less than two (2) hectares. This study aimed to make a recommendation model using association rules to suggest recommendations for dairy farm management to smallholder farmers based on their farm details. The key objectives were to identify requirements of the recommendation model, to develop the recommendation model, to deploy and validate the model as an end user mobile tool. The study was conducted in Arusha and Kilimanjaro regions in Tanzania. Systematic random sampling was used for data collection, Apriori algorithm was used in recommendation model development and incremental methodology was used in mobile application development. A recommender system was developed through association rules mining to provide recommendations to the smallholder dairy farmers that help them to increase their production using strategies available from fellow farmers in the same cluster. These association rules were used to help small holder dairy farmers to move from low milk production (9.15  3.25 litres) to medium milk production (11.08  4.29 litres) and then to high milk production (14.45  5.12 litres) gradually. The Dairy Recommender System mobile application with a recommendation model deployed in the backend was developed for farmers to easily get information on how to improve dairy farming practices.
Development of a web-based record management system for primary cooperative societies in Tanzania: a case of Kilimanjaro region
(NM-AIST, 2024-08) Germinous, George
Co-operative societies in Tanzania play a vital role in the country's economic development, serving as a means to combat poverty and enhance members' economic prospects. However, the current in-house methods of documentation and record-keeping in co-operatives poses challenges for stakeholders such as District Co-operative Offices (DCOs), Regional Co- operative Offices (RCOs), the registrar of co-operatives, and researchers. Manual processes or limited-access summary reports hinder real-time data availability for management and decision making. This study addressed these challenges by developing a Centralized Record Management System (CRMS) for District Co-operative Offices and their members. To collect user requirements, administered questionnaire was used among District Co-operative Offices (DCOs). the Registrar of cooperatives and researchers in the co-operative arena. The study revealed that both hard and soft copies were used for record’s keeping purposes interchangeably. Physical files such as clip and spring files, as well as file cabinets, were commonly used for storing hard copies, while electronic files in various formats (e.g., word processors, spreadsheets, pdf, and images) were employed for digital record-keeping. Both manual tools (pens and papers) and electronic tools (computers, smartphones, and cameras) were used for data processing. Following careful analysis of user requirements, CRMS was designed and implemented using a Joint Application Development Model incorporating an Evolutionary Prototyping (EP). The developed CRMS facilitates online registration and approval of cooperative societies, reporting of cooperative data to the DCO, DRCO and Registrar, and disseminate the generated cooperative reports to the stakeholders. By implementing the CRMS, primary co-operative societies can benefit from streamlined processes and improved access to data, ultimately fostering their growth and contributing to a more transparence and thriving co-operative sector.
Development of an accessible augmented reality application to enhance indoor navigation: a case of higher learning institutions
(NM-AIST, 2024-08) Samson, Frank
This study focused on developing an accessible Augmented Reality (AR) application to enhance indoor navigation, using a case study of learning institutions. A case of The Nelson Mandela African Institution of Science and Technology and Baden-Wuerttemberg Cooperative State University institutions were selected for this study. The project’s objectives included gathering requirements for the development and understanding of the challenges faced by students, staff, and visitors who can be visually or non-visually impaired while navigating indoors in unfamiliar buildings. The project's significance lies in its potential to enhance indoor mobility, and accessibility, save time, and foster innovation in education technology. The application aims to create a more inclusive learning environment by improving indoor navigation. Primary data were collected through interviews and focus group discussions, creating a foundation requirement for developing the application. The study used Scrum Agile methodology to develop the application and web portal from the collected user requirements. The application was designed to run on smartphones and further, a web-based portal was designed for staff to update their visitor availability status. Development tools such as Unity, Microsoft Visual Studio, PHP, Python, and APIs were used. This study highlights the challenge of indoor navigation in higher learning institutions and the limited effectiveness of current solutions, especially for people with disabilities particularly the visually impaired. Leveraging AR technology, known for enhancing real-world perception by overlaying virtual content on top of users' real-world views, a user-friendly and accessible indoor navigation application was developed. 80% of the results demonstrated the effectiveness and usability of the application, with positive feedback from users highlighting its potential to transform the indoor navigation experience. In conclusion, the study underscores the need for further research to enhance indoor navigation, localization, and accessibility for a diverse range of users and use cases. i
A deep learning model for early detection of maize diseases in Tanzania
(NM-AIST, 2024-08) Mayo, Flavia
Agriculture is considered the backbone of Tanzania’s economy, with more than 60% of residents depending on it for their livelihood. Maize is the main and dominating food crop in the country, accounting for 45% of all farmland produce. Inevitably, its production is affected by diseases that, if detected early, could be easily treated. Maize Streak Virus (MSV) and Maize Lethal Necrosis (MLN) are commonly reported diseases that are often detected too late by farmers. This raised the need for a sophisticated method for early detection of these diseases to enable timely treatment. This study investigated the potential of developing a deep-learning model for the early detection of maize diseases in Tanzania. Imagery datasets were used for model training, they were collected through physical observation with the help of a plant pathologist from three regions of the country; Arusha, Kilimanjaro, and Manyara. Two models, a Convolutional Neural Network (CNN) and a Vision Transformer (ViT), were developed from scratch and classified into four classes: Healthy, MLN, MSV, and WRONG. The results revealed that the ViT model outperformed the CNN model, achieving validation accuracies of 0.931 and 0.9096 respectively. Despite the superior performance of the ViT model, the CNN model was selected for deployment in a mobile-based application due to its smaller size, lower memory, and small computational requirements. A quantitative research method using a survey questionnaire was employed to gather requirements for system development and validation. Validation of the model's performance through questionnaires administered to farmers and agricultural experts, yielding highly positive feedback. This enthusiastic reception highlights the potential impact of the developed application on improving early disease detection and enhancing maize productivity in Tanzania. Empowering stakeholders of agriculture to identify more effective methods of managing them before serious harm is done. Consequently, custodians of agriculture, such as the Ministry of Agriculture, organizations, and other companies, will have the potential to detect maize diseases early, henceforth adding up to the increase in the Gross Domestic Product (GDP) and guaranteeing the country’s food security.
Image segmentation deep learning model used to Identify black Saigatoka And Fusarium wilt in banana early
(NM-AIST, 2024-05) Elinisa, Christina
Bananas are among the most widely produced perennial fruit crops. Farmers largely produce bananas because they are important staple food and cash crops. However, bananas are highly affected by Fusarium Wilt and Black Sigatoka diseases. These diseases cause yield losses ranging from 30% to 100% of all the banana produce. Farmers face challenges in detecting and mitigating the effects of these two banana diseases because of a lack of knowledge of the diseases and the use of traditional eye observation method in detection. This study is inspired by the success of deep learning and computer vision in detecting a wide range of plant diseases. The study proposed the use of deep learning to automate the early detection of Fusarium Wilt and Black Sigatoka banana diseases. Mask R-CNN and U-Net image segmentation deep learning models were assessed for instance and semantic image segmentation. A dataset comprising 27 360 images of banana leaves and stalks that are healthy, Fusarium Wilt infected, and Black Sigatoka infected collected from the farm was used to train the models. An addition of 407 images of other things apart from the banana plant were downloaded from the internet and used to train the CNN model. From the experiments, the Mask R-CNN model achieved a mean Average Precision of 0.045 29 in segmenting the two banana diseases. The U-Net model achieved an Intersection over Union (IoU) of 93.23% and a Dice Coefficient of 96.45%. Similar results were obtained by Loyani et al. (2021) when they segmented a tomato plant paste called tuta absoluta using a U-Net model. Their model achieved a Dice Coefficient of 82.86% and an Intersection over Union of 78.60%. Additionally, the Fusarium Wilt and Black Sigatoka infected banana leaves and stalks were segmented using the Mask R-CNN and U-Net models. The CNN model yielded an accuracy of 91.71% in classifying the two banana diseases. Similar results were obtained by Sanga et al. (2020) when they deployed an Inceptionv3 model, which achieved an accuracy of 95.41%. The CNN model was deployed in a mobile application to be used by farmers to detect the two banana diseases early. The mobile application could detect banana diseases early and provide research-based mitigation recommendations that smallholder farmers and other agricultural stakeholders can use to avoid yield losses and financial losses.
Development of data-driven web-based system for regional programme accreditation: a case study of the inter-university council for East Africa (IUCEA)
(NM-AIST, 2024-08) Nkeshimana, Carmel
In response to evolving regional program accreditation requirements, a user-centric Regional Program Accreditation System (RPAS) was meticulously designed to cater to the diverse needs of accreditation bodies, educational institutions, and expert reviewers. The development journey commenced with an extensive requirement-gathering phase, engaging end-users to shape the system's design and functionality. RPAS was built using cutting-edge technologies, including ReactJs, Django, API integration, Tailwind CSS, and MySQL, ensuring scalability, security, and flexibility. Agile methodologies, particularly Extreme Programming (XP), were embraced, enabling expedited development and fostering close collaboration, ensuring adaptability to users' evolving needs. The RPAS streamlines the accreditation process with a user-friendly interface and real-time collaboration features, thereby enhancing transparency and operational efficiency. To guarantee data security, reliable authentication procedures to guarantee data security, reliable authentication procedures have been implemented. Additional user-friendly features, such as night mode, QR code-based certificate verification, and automated notifications, were introduced for convenience. The project prioritized user training and support, data privacy, and seamless user feedback integration to make RPAS an intuitive and responsive tool. It is noteworthy that the Rasch model was utilized in the development of a similar system, but its application was limited to primary and secondary schools in Indonesia. These enhancements further empower RPAS, making it a valuable tool for streamlining accreditation processes and contributing to the elevation of program quality and educational excellence in the region. The RPAS represents a significant advancement in the accreditation landscape, offering an integrated, user-centric solution that addresses the unique needs of all stakeholders involved in the accreditation process. Its potential for further improvements makes it a promising instrument for the future of educational quality assurance in the region.
An enhanced IoT-based wristband for remote monitoring and early detection of hypertension complications in Uganda
(NM-AIST, 2024-08) Nkoloogi, Blasius
Remote monitoring systems can transform healthcare for non-communicable diseases like hypertension. Despite widespread blood pressure testing, real-time communication and record storage remain challenging in Uganda. This project developed an enhanced Internet of Things (IoT) based wristband for remote monitoring and early detection of hypertension complications. The system integrates a wearable wristband and web application to track vital signs blood pressure, heart rate, oxygen saturation, and body temperature transmitting real-time data wirelessly. It alerts patients, next of kin, and medical practitioners to critical hypertension levels, enabling early intervention against heart disease, stroke, and kidney disease. Data collection involved 243 patients and 17 medical practitioners from Rocket Health Clinic in Uganda. Qualitative methods included focus groups, observations, and analysis of Electronic Medical Records (EMR), while quantitative methods utilized patient surveys. Additional data from reports, journals, books, databases, and websites on blood pressure monitoring systems were also analyzed using Power BI. The development followed Extreme Programming (XP) agile methodology, accommodating evolving requirements under tight deadlines. Validation indicated the system is easy to use, accurate, and reliable. The wristband hardware includes the DOIT ESP32 DevKit v1 microcontroller, MKB0803 blood pressure sensor, MAX30102 pulse oximeter sensor, DS18B20 body temperature sensor, OLED SSD1306 display module, UC15 3G module, and SIM800L GSM module. The web application was developed using Laravel, Vue, Inertia, and EChart. Future work aims to enhance system accessibility regardless of smartphone or internet access, potentially through voice-based interfaces. Evaluations will extend to managing chronic diseases like diabetes, leveraging insights to improve healthcare in resource-limited settings.

Browse

Recent Submissions