Department of Software Engineering

Permanent URI for this collectionhttps://repository.mu.edu.et/handle/123456789/1157

Browse

Search Results

Now showing 1 - 3 of 3
  • Item
    Integrating Ethereum Blockchain and GraphQL for a Secure Graduate Verification System at Mekelle University
    (Mekelle University, 2026-01-28) Tilahun Mamuye Gidey
    Ensuring the integrity and efficiency of academic record verification has become increasingly important for modern educational institutions. This study presents a blockchain-powered verification system specifically designed for confirming the credentials of graduated students from Mekelle University. By integrating Ethereum blockchain with GraphQL APIs, the system enhances transparency and reliability in the verification process. The university’s existing system, built with Ruby on Rails, lacked automated verification, relied heavily on centralized control, and was prone to delays and potential data tampering. To overcome these issues, a decentralized application (DApp) was developed using various tools, including Ethers.js, Node.js, Ganache, Apollo Server, GraphQL, and React. This application enables the secure submission and retrieval of student records through Ethereum smart contracts. Data can be uploaded via CSV files or manually entered through forms, and each record is retrievable using a unique student ID, ensuring data immutability and public verifiability. Stakeholder feedback was gathered through interviews, and thematic analysis was used to assess the system’s usability, scalability, and trustworthiness. Findings showed strong support for the blockchain-based system, with over 90% of participants agreeing that it improves transparency and reduces the risk of credential fraud. This research demonstrates a feasible bridge between traditional university information systems and decentralized technologies, highlighting both the practicality and institutional readiness for adopting blockchain in higher education.
  • Item
    Develop Deep Learning-Based Face Recognition Model for Masked and Unmasked Faces
    (Mekelle University, 2025-10-14) Tirhas Kebede
    Traditional facial recognition systems are no longer as effective due to the widespread use of face masks. To overcome this difficulty, the study suggests a facial recognition (FR) system based on a unified framework specifically designed for partial facial occlusion scenarios that can reliably identify people wearing or not wearing face masks. It is built on the Inception ResNet V1 deep learning model and was trained using the Real-World Masked Face Dataset (RMFD) with 525 subjects. The proposed system outperforms existing methods like FaceNet, achieving a test accuracy of 98.3% through rigorous subject-level cross-validation, while maintaining real-time performance at 45 FPS with optimized model size of 96MB. The solution, which is optimized for real-time deployment, can be used for public health monitoring, access control, and security verification. In both masked and unmasked faces, experimental results show strong generalization and consistent accuracy, allowing for a smooth integration into safety-critical applications without sacrificing effectiveness.
  • Item
    Development of a Hybrid Pretrained Deep Learning Model with Explainable AI for Tomato Disease Detection
    (Mekelle University, 2025-11-18) Abrhaley Gebreslassie
    Tomato cultivation plays a pivotal role in ensuring global food security and sustaining national economies. However, tomato crops are frequently threatened by a variety of diseases caused by bacterial, viral, and fungal pathogens, leading to significant yield losses. Conventional disease identification approaches, which primarily rely on manual inspection by agricultural experts, are time-consuming, error-prone, and inaccessible to smallholder farmers, particularly in resource-limited settings such as rural areas of Tigray situated in the Northern Part of Ethiopia. This thesis proposes a hybrid deep learning model augmented with Explainable Artificial Intelligence (XAI) techniques to enhance the accuracy, interpretability, and practical applicability of automated tomato disease detection systems. A benchmark dataset comprising 17,920 annotated images representing ten distinct tomato disease classes was utilized for training and evaluation purposes. Two pre-trained convolutional neural network (CNN) architectures, namely ResNet-50 and MobileNetV2, were selected for comparative analysis based on their performance and computational efficiency. Based on the experimental results of the proposed method, the ResNet-50 model achieved a training accuracy of 99.25% and a validation accuracy of 94%, whereas MobileNetV2 attained 92.6% training and 87.4% validation accuracy. To improve the generalization capabilities of the models and mitigate overfitting, several data augmentation strategies, including rotation, flipping, and scaling, were employed. Additionally, Gradient-weighted Class Activation Mapping (Grad-CAM) was applied to visualize the class-discriminative regions within input images to provide interpretability and revealing potential biases in model predictions. The results demonstrate that the integration of deep learning with XAI techniques yields an effective and transparent solution for tomato disease detection. The proposed approach offers valuable insights for deploying intelligent diagnostic tools in precision agriculture, particularly benefiting smallholder farmers by facilitating early disease detection and improved crop management.