Scholarly publications produced by the members of University of Ruhuna http://ir.lib.ruh.ac.lk/handle/iruor/8 2026-09-04T10:44:27Z 2026-09-04T10:44:27Z Gunawickrama, S.H.K.K. Bandara, U.G.R.S. Dissanayaka, D.M.S.C. Dissanayake, D.M.M.I.T. Perera, H.L.D.U.G. http://ir.lib.ruh.ac.lk/handle/iruor/21723 2026-09-04T09:56:19Z 2026-03-04T00:00:00Z

High-Volume Automated MCQ Marking: An Image Processing Approach for Enhanced Efficiency and Accuracy. Gunawickrama, S.H.K.K.; Bandara, U.G.R.S.; Dissanayaka, D.M.S.C.; Dissanayake, D.M.M.I.T.; Perera, H.L.D.U.G. This study proposes the development of an AI-powered optical mark recognition (OMR) system that aims to automate and enhance the grading of multiple-choice question (MCQ) answer sheets. Traditional OMR systems rely on shaded bubbles or pencil marks, which can cause recognition errors, especially when students use different marking styles, such as crosses instead of shading. The manual marking process can take a considerable amount of time for large numbers of students. To overcome these limitations, the proposed system combines advanced image processing techniques with a Convolutional Neural Network (CNN) to accurately classify bubbles as crossed, shaded, or empty. A mathematically calculated and defined A5 bubble-sheet with a 50-question layout was designed to support two paper structures, enabling shuffled question papers for enhanced exam security. These structures make the system more practical, user-friendly, and suitable for large-scale academic and institutional assessments. The system was implemented using Python with PySide6 for the graphical user interface. In addition, OpenCV, TensorFlow, NumPy, and Pandas were used for processing and model training. A custom dataset was developed using real student-marked papers representing different marking behaviors and image conditions. The final CNN model achieved a validation accuracy of 98.87%. The model demonstrated strong generalization and consistent performance across different lighting and image quality conditions. Furthermore, the Gemini Pro API was incorporated to automatically extract student registration numbers from scanned answer sheets, ensuring accurate mapping between responses and student identities. The desktop application provides features for automated scoring, manual verification by examiners, and exporting results in Excel format. It also supports evaluating shuffled MCQ papers, enabling examiners to mark answer sheets that belong to different paper versions with different question orders. Additionally, the system includes registration number detection. The system substantially reduces grading time while maintaining human-verified reliability through dual examiner review features. The solution provides a practical, scalable, and cost-efficient alternative to commercial OMR systems for large-scale assessments in universities.

2026-03-04T00:00:00Z Hettiarachchi, H.P.P.P. Kavindya, P.P. Jayasinghe, D.M.S.N. Kumarasiri, L.I.N. Sandamali, G.G.N. http://ir.lib.ruh.ac.lk/handle/iruor/21722 2026-09-04T09:24:50Z 2026-03-04T00:00:00Z

AI-Based Real-Time Multi-Agent Depression Detection and Therapy System. Hettiarachchi, H.P.P.P.; Kavindya, P.P.; Jayasinghe, D.M.S.N.; Kumarasiri, L.I.N.; Sandamali, G.G.N. Depression is a prevalent mental health disorder in the world, but diagnosis is constrained by stigma, cost and lack of clinical services. This study describes an AI-Based Real-Time Multi-Agent Depression Detection and Therapy System which aims at early detection and provides therapy suggestions. NLP and speech emotion analysis are incorporated to assess the user input and calculate the level of depression using the Patient Health Questionnaire-9 (PHQ-9). Five specialized agents work collaboratively to process text and voice data, retrieve linguistic and emotional features, and categorize mental states from minimal to severe. A fusion model of PHQ-9 scores, classifier confidence, and speech-based emotion indicators produces a factor-calibrated depression index for precise severity differentiation. Depending on the identified level, the system can offer AI-assisted therapeutic care for minimal or moderate cases, while severe cases are referred to professional care. For voice-based emotion detection, wav2vec 2.0 was trained and reached 81% testing accuracy when classifying emotions into happy, sad, fearful, angry, and neutral categories. Among the 10 participants diagnosed with depression, the system correctly matched the actual severity level for 9 individuals, resulting in an accuracy of 90%. Among these matched cases, a comparison based only on PHQ-9 severity revealed one mismatch: PHQ-9 indicated minimal severity, while both the actual diagnosis and the system identified a moderate level. This demonstrates that the proposed system provides a more accurate severity classification than reliance on PHQ-9 scores alone. A comparison of Large Language Models showed that GPT-3.5 and Mistral produced the most fluent outputs, with BERTScores of 71% and 69.2% respectively, outperforming Gemini at 61% and Claude at 59.8%. Among them, GPT-3.5 was identified as the most suitable model to integrate into the system due to its overall performance.

2026-03-04T00:00:00Z Roshani, K.W.H. Agakar, C. Laksayan, T. Mathushagan, J. Wijesinghe, B.G.S.U. Liyanagamage, L.I.M. Weerasinghe, T.N. Priyankara, W.N.B.A.G. Senevirathne, C.K.W. http://ir.lib.ruh.ac.lk/handle/iruor/21721 2026-09-04T07:20:34Z 2026-03-04T00:00:00Z

Mitigation of Atmospheric Duct Interference (ADI) in LTE-TDD Networks: A Machine Learning-Based Approach. Roshani, K.W.H.; Agakar, C.; Laksayan, T.; Mathushagan, J.; Wijesinghe, B.G.S.U.; Liyanagamage, L.I.M.; Weerasinghe, T.N.; Priyankara, W.N.B.A.G.; Senevirathne, C.K.W. Atmospheric Duct Interference (ADI) causes abnormal long-range radio propagation due to refractivity inversions in the lower atmosphere, resulting in severe uplink degradation in LTE-TDD cellular deployments. This research proposes a data-driven framework to automatically detect, predict, and mitigate ADI while preserving network coverage and uplink quality. The proposed framework was developed using two heterogeneous data sources. Network-side measurement data were obtained from Dialog Axiata PLC, collected from LTE-TDD deployments in the Jaffna Peninsula, with approximately 100,000 time-indexed records from over 580 cells across 111 base stations, including uplink interference statistics, symbolwise uplink power measurements (UL0-UL13), antenna configuration parameters, and geographical coordinates. Environmental and atmospheric parameters were obtained from the Copernicus Climate Data Store, including temperature, pressure, and humidity profiles used to characterize ducting conditions. After preprocessing and labeling, numerical and categorical features were standardized and encoded using a Python-based pipeline. Class imbalance was addressed using random undersampling, and the data were split into 70% training, 15% validation, and 15% testing sets. Three supervised classifiers—Random Forest, Support Vector Machine, and k-Nearest Neighbors—were trained and evaluated using accuracy, confusion matrices, and cross-validation. Among the evaluated models, the Random Forest classifier demonstrated the most stable performance, achieving the highest test accuracy of 97.8% in distinguishing normal and interferenceaffected conditions. Feature analysis confirmed that uplink interference levels, symbol-wise power ramping behavior, and spatial indicators are dominant predictors. Upon ADI detection, a coordinated mitigation strategy involving adaptive antenna down-tilting, transmit power adjustments, and coverage compensation from neighboring cells was applied using radio propagation modeling. Simulation results demonstrate that over 70% of ADI-induced coverage loss can be recovered without introducing additional spectrum usage. This research presents an original, scalable solution for intelligent ADI management in LTE-TDD networks and provides a practical foundation for future ADI-aware optimization.

2026-03-04T00:00:00Z Kaluarachchi, K.A.S.D. Senanayaka, S.U. Senevirathne, S.M.K.H.T. Sandamali1, G.G.N. http://ir.lib.ruh.ac.lk/handle/iruor/21699 2026-09-03T09:43:01Z 2026-03-04T00:00:00Z

Integrated Health Monitoring System for Preventive Care and Doctor-Patient Interaction. Kaluarachchi, K.A.S.D.; Senanayaka, S.U.; Senevirathne, S.M.K.H.T.; Sandamali1, G.G.N. Diabetes, hypertension, and cholesterol disorders are some of the major chronic diseases globally. Early diagnosis, continuous monitoring, and strong doctorpatient communication are essential for controlling these diseases. While existing digital health systems focus on either medical data or lifestyle monitoring, they lack an integrated approach for preventive care. This system proposes a personalized health monitoring system combining medical data tracking, nutritional monitoring, and interactive doctor-patient communication. The system is built using Node.js, React/React Native, and Firebase, enabling real-time data management. The system has two types of users: patient and doctor. Patients can enter or upload their medical records into the system, including details related to blood pressure, blood sugar, lipid profile, and full blood count. They can monitor their dietary intake manually or by uploading pictures of consumed food. A convolutional neural network trained on 5,000 labeled food images across five food classes was used to recognize food items and estimate nutritional parameters such as calories, sugar, and cholesterol, achieving a validation accuracy of approximately 90%. The estimated values are compared with recommended daily intake limits, and warning messages are generated when thresholds are exceeded, providing actionable insights to support healthier dietary choices. A dedicated dashboard enables doctors to view patient medical histories and provide feedback. The system supports active patient engagement through continuous selfmonitoring and timely remote medical feedback. By enabling early identification of potential health issues and remote consultations, the system reduces unnecessary hospital visits, while the integrated real-time chat platform facilitates efficient doctor-patient communication. Overall, the system supports personalized health monitoring and preventive healthcare through integrated data tracking and interactive digital communication.

2026-03-04T00:00:00Z