MPhil Thesis
Browse
33 results
Search Results
Item Rice Leaf Disease Detection Using Machine Learning Technique(Daffodil International University, 2024-07-13) Hasan, Md MehediThis study explores the application of deep learning models for the detection of rice leaf diseases, a critical issue impacting global rice production and food security. The research focuses on five advanced deep learning architectures: Convolutional Neural Network (CNN), Xception, VGG19, MobileNetV2, and InceptionResNetV2. Utilizing a dataset comprising 6,420 images across four disease categories—Brown Spot, Tungro, Bacterial Blight, and Blast—each model was trained and evaluated to determine its accuracy and effectiveness in disease classification. The proposed methodology encompasses data collection, labeling, image processing, model selection, training, evaluation, and testing. Results demonstrated that the CNN model achieved the highest accuracy at 98.44%, followed closely by MobileNetV2 at 97.82%, VGG19 at 96.57%, InceptionResNetV2 at 95.43%, and Xception at 95.07%. These high accuracies underscore the potential of deep learning models in early disease detection, which is crucial for timely intervention and effective crop management. Comparative analysis with traditional machine learning approaches such as Support Vector Machines (SVM) and Decision Trees, which typically yielded lower accuracies between 81.8% and 97%, highlights the superior performance of deep learning techniques. Furthermore, the study discusses the ethical considerations, including data privacy, accessibility for small-scale farmers, and the need for unbiased models, ensuring equitable benefits across diverse agricultural contexts.Item Enhancing guava harvest forecasting in Bangladesh through supervised machine learning models(Daffodil International University, 2024-01-01) Munia, Sahela KhanAccurate forecasting of guava harvest is essential for efficient resource allocation, market planning, and mitigating post-harvest losses. In Bangladesh, the guava industry faces challenges in predicting harvest yields due to the complex interaction of various environmental factors. This study proposes a novel approach to enhance guava harvest forecasting in Bangladesh through the application of supervised ML models. The research leverages historical guava production data and corresponding meteorological variables, including temperature, humidity, precipitation, and solar radiation. These variables are used as input features for training and testing several supervised ML models, such as linear regression, decision trees, random forests, support vector machines, and artificial neural networks. A comprehensive dataset comprising guava production records and meteorological data from multiple regions in Bangladesh is collected and preprocessed. Feature engineering techniques are employed to extract relevant information from the data and optimize model performance. The dataset is then divided into training and testing sets for model development and evaluation. Performance metrics such as MAE, RMSE, MSE are used to assess the accuracy and reliability of the machine learning models. Where the highest accuracy 84.72% is achieved by DTR. And the lowest accuracy is achieved by LinR accuracy of 43.07%. The models' forecasting capabilities are compared, and the most effective model is identified. The results demonstrate that the supervised machine learning models exhibit promising performance in guava harvest forecasting, outperforming traditional statistical methods. The selected model achieves high accuracy and provides valuable insights into the influence of meteorological variables on guava production. The findings of this study have significant implications for the guava industry in Bangladesh, helping to enhance productivity, reduce wastage, and promote sustainable agricultural practices. Moreover, the methodology presented are extended to other regions and crops, facilitating improved harvest forecasting in diverse agricultural contexts.Item Brain tumor detection from MRI medical images based on machine learning algorithms(Daffodil International University, 2023-12-21) Omy, Jannatul FariaThis study aims to develop an efficient and accurate system for the early detection of brain tumors using machine learning algorithms applied to magnetic resonance imaging (MRI) medical images. Brain tumor occurs because of anomalous development of cells. It is one of the major causes of death in adults around the globe. Millions of deaths can be prevented through early detection of brain tumors. Earlier brain tumor detection using Magnetic Resonance Imaging (MRI) may increase a patient's survival rate.machine learning) has gained prominence in almost every field where decision-making is involved in recent years, spanning economics, health care, marketing, and sales. In the field of healthcare, machine learning & deep learning have shown promising results in a variety of fields, namely disease diagnosis with medical imaging, surgical robots, and boosting hospital performance. One such application of deep learning to detect brain tumors from MRI scan images. In MRI, tumor is shown more clearly that helps in the process of further treatment. This work aims to detect tumors at an early phase. A comprehensive dataset of MRI scans, encompassing both tumor and non-tumor cases, is utilized to train and validate the proposed machine learning models. Preprocessing techniques, including image enhancement and normalization, are applied to standardize the input data. Various machine learning algorithms, such as convolutional neural networks (CNNs), MobileNet model with ImageNet weights from keras and decision trees, are implemented and compared to identify the most effective approach for brain tumor detection.In this research of brain tumor classification, using machine learning, and built a binary classifier to detect brain tumors from MRI scan images. The classifier used transfer learning and obtained an accuracy of 96.5% and visualized the model’s overall performance.The presents a model which is based on machine learning algorithms to detect brain tumors from magnetic resonance images with high accuracy. A Convolutional Neural Network (CNN) has been used as the algorithm for feature extraction, and segmentation. The dataset used has been acquired from kaggle.Item Paddy Yield Estimation By Deep Learning Approach(Daffodil International University, 2024-01-29) Islam, AshrafulThree crucial crops in Bangladesh are farmed concurrently based on the country's climate and seasons. The individuals mentioned are Aush, Aman, and Boro. Various sorts of damage occur in Bangladesh at different times as a result of natural catastrophes, in accordance with the country's climate. Examples include storms, torrential downpours, floods, and river overflows. They inflict harm. Rice is the primary agricultural product of Bangladesh. The rice yield is significantly impacted by these natural disasters. Consequently, a multitude of different natural disasters transpire. These encompass agricultural yield decline, insufficiency in food supply, and potentially even widespread starvation. In order to address these issues, I have devised a model that can accurately predict the rice yield for the current season by examining historical data. For this particular situation, I have employed D planning. Among the three models I have tested, the LSTM model demonstrated superior performance. The third model demonstrated an R2 square A score of 0.77% with less of loss at 0.22%. Here use of 1560 data. The model has achieved unprecedented advancements.Item Machine Learning Approach to Predict Tomato Leaf Disease(Daffodil International University, 2023-09-02) Akter, ShumayaThe machine learning approach used in this thesis uses a large dataset of leaf photos and associated disease labels to predict tomato leaf diseases. A large number of tomato leaf photos, comprising both healthy and diseased leaves, is gathered for the research. Techniques for preprocessing are used to enhance picture quality and extract relevant characteristics. The objective of the technique is to find distinguishing traits in leaf pictures that distinguish between various disease classifications. Convolutional neural networks (CNNs), decision trees, random forests, support vector machine, and other machine learning methods are assessed for disease prediction. To evaluate the performance of the models, the cross-validation methods are used to verify them on the labelled dataset. To learn more about the visual patterns connected to each condition, feature importance analysis is done. Techniques for transfer learning are investigated to make better use of taught models. The experimental findings show a high degree of prediction accuracy for tomato leaf diseases, with transfer learning-based CNN-based models outperforming more conventional methods. The research helps create an automated method for early disease identification, which helps farmers and specialists execute effective disease control techniques on time, reducing crop losses, and maintaining sustainable tomato production. The paper highlights the potential of machine learning in plant pathology and encourages further investigation into related methodologies for additional plant species and disease classes.Item Higher Education Student's Performance Evaluation Using Machine Learning Techniques(Daffodil International University, 23-03-01) Efty, MD.Kamrul HasanFailure and success in the classroom have real-world implications for achieving economic success in the knowledge-based economy. Using early detection markers (such as age, reading frequency, and CGPA), this research aims to forecast the likelihood of students' academic performance in order to provide prompt and effective remediation. On the basis of secondary data acquired from students' information systems, a machine learning approach was employed to create a model. In this paper, our main aim is to predict student performance for 3 specific factors student scientific book reading frequency, extra work conditions, and weekly study time. So we are using five machine learning algorithms KNN, Random forest, Decision tree, Linear regression, and GBC, and also use almost 1200 student attribute datasets. For students with extra work conditions random forest algorithms given the highest 99 % accuracy. For student scientific book reading frequency random forest and decision tree are given the highest 98 % accuracy. For students weekly study hours random forest and KNN given highest 97 % accuracy.Item Paddy Leaf Disease Detection Using Machine Learning Technique(Daffodil International University, 23-02-18) Konik, Mustafizur Rahman; Swapnil, Azaz AhmadOne of the most important key resources to prevent global warming on the planet is plants. But the plants are suffering from various diseases. In recent time, research has been begun for acknowledgement of plant disease. Paddy disease detection is the key intention of this paper. Brown Spot Disease (BSD), Leaf Blast Disease (BD), and Leaf Blight Disease (LBD) are a few of the paddy diseases that prevent the paddy from growing and protecting every portion of the plant including diseases that can affect paddy at various stage of growth. This research examined 3 different disease kinds as well as one group of healthy paddy leaves. Bacteria, fungi, and other organisms are among those that can cause paddy disease. The Technique was created to eliminate noise automatically by decreasing the time needed to measure the impact of paddy leaf disease on humans so using machine learning techniques k-means for image segmentation and an automated detection method to get the best results for finding paddy leaf disease with the approach of machine learning using classifications with the best accuracy. To measure classification of this paper K-Fold cross validation techniques has been used. Applying 4 classes of paddy leaf’s into Random Forest, Decision Tree, Logistic Regression and SVM like support vector classifier (SVC), among then Random Forest gave the highest 94.16% accuracy with the using of K-fold cross validation techniques in predicting the three classes of paddy leaf disease with one group of healthy paddy leaves.Item A Machine Learning-Based Technique for Predicting Heart Disease(Daffodil International University, 23-02-12) Nahin, Mahmudur Rahman; Shawon, SadikuzzamanPhysical diseases including heart disease have been on the rise recently. The subject is well-known in the modern world. The majority of individuals have an issue with heart disease. The discrepancies between the normal and afflicted diagnosis report ratios serve as a gauge of the condition. Heart illness is a condition that has been the subject of several investigations in the past. We have identified a few excellent chances to develop the methodology. We suggest employing efficient algorithm models to forecast dangers and raise early awareness. Our suggested approach is suited for straightforward heart disease predictions and is simple to apply in the actual world. The Kaggle website hosted the dataset. In our model, we have implemented some different classifiers named Random Forest (RF), Logistic Regression (LR), Gradient Boosting (GB), Support Vector Classifier (SVC), Adaboost Classifier (ABC), Naïve Bayes (NB), Decision Tree (DT) algorithms. Random Forest (RF) given an accuracy of 90.22%, Logistic Regression (LR) given accuracy of 89.67%, Gradient Boosting (GB) given accuracy of 89.67%, Support Vector Classifier (SVC) given accuracy of 91.85%, Adaboost Classifier (ABC) given the accuracy 91.30%, Naïve Bayes (NB) given the accuracy 89.67%, Decision Tree (DT) given the accuracy 91.85%. We have used ensemble techniques to get the best accuracy. Our voting classifier RDSGLGA gave the best accuracy of 93.478%. Another voting classifier RDS gave an accuracy of 92.39%. To assign the optimal parameters to each classifier, we employed hyperparameter tuning. The experimental investigation reviewed the results of previous recent studies and found that RDSGLGA performed best, with an accuracy rate of 93.478% in terms of making heart disease predictions.Item Bangla News Article Categorization Using Machine Learning(Daffodil International University, 23-02-12) Haque, MD Al Shahriar; Shawda, UmmeBangla language got familiar many years ago in the world and Many online Bangla news portals are growing day by day. We can get news within a few seconds with their help of them. Some media are telecasting news by live stream and some are publishing news through online news portals. With their help of them, much news is being published day by day. We are familiar with many new things by seeing/reading the news. This news is not separated by its specific categories the problem arrives because every people don’t like every category. For this reason, they feel disturbed to read the news but very few researchers are working in Bangla news and at this time data gap is increasing very rapidly. In this paper, we try to solve this problem by Machine learning. we collect data by the web crawler. Our dataset has 408470 rows and collects data 120 thousand. We use label mapping for category labeling and to get sequence we use a tokenizer, for data preprocessing we use a slicer to get the same sample in every category. We use flatten, embedding, and dense, and we use ‘adam’ optimizer, for loss function ‘sparse categorical cross entropy’, for visualizations we use a heatmap, and confusion matrix, for classification we use some classifiers like SVM, KNN, decision tree, random forest, naive bayes, Gradient Boosting Classifier. After using the decision tree, and random forest we get a training accuracy is 98.08%.Item Heart Disease Prediction Using Machine Learning:(Daffodil International University, 23-02-18) Hossin, MD. Murad; Bhuiyan, MD. RifatHeart disease is one of the main causes of death worldwide and the most dangerous ailment. Early identification of cardiovascular disease will reduce mortality. The medical establishment has struggled in recent years to accurately anticipate cardiac disease. According to recent data, one person dies every minute from heart disease. Data science is needed to comprehend the vast volumes of new healthcare data. KNN, LR, AdaBoost, XGB, RF, GB, SVM, and DT machine-learning algorithms are used to forecast cardiac disease. Using these algorithms, we could analyze a person's heart disease risk based on dataset attributes. This study used two types of data. The first heart disease dataset had 918 patient records, 11 attributes, and one target. This dataset combines five well-known cardiac datasets. The second dataset on cardiovascular disease included 70000 patient records, 11 characteristics, and a single goal. This research offers a comparison study by investigating the efficacy of numerous machine learning methods. For our first and second datasets, Gradient Boost (GB) was the most accurate, with 91.80% and 74.50%, respectively. Considering the results of the trial, the Gradient Boost (GB) algorithm has the highest level of accuracy, which is 91.80%, compared to other models and studies being done at the time. A realistic web application is also developed.