Explore student projects
Search and filter published projects from across departments.
Design and Implementation Of An AI-Based Itsekiri Sign Language Translator
This study explores deep learning approaches for sign language recognition. It investigates existing methodologies, architecture performance, and gaps in current literature, while evaluating theoretical frameworks for accessible design to bridge communication barriers.
Design and Implementation of CodeQuest: An Interactive, Multi-Language Coding Learning Platform for Children
Programming has become a basic literacy of the digital age, yet many children struggle to learn it because it is often introduced through dense syntax, abstract vocabulary and unforgiving error messages. Children who start with visual, block-based tools usually find the early concepts approachable, but a second problem appears later: the jump from dragging blocks to typing real code in a professional language is abrupt, and many learners lose confidence at exactly that point. This project presents the design and implementation of CodeQuest, an interactive, gamified coding platform that addresses both problems within a single, self-contained web application. CodeQuest is a full-stack TypeScript application. The frontend is built with React 18.3, Vite 6 and React Router 6.28, while the backend runs on Node.js 24 with Express 4.21, JSON Web Token (JWT) authentication, bcrypt password hashing and role-based access control. A dual database adapter uses SQLite through the Node.js DatabaseSync module during local development and switches automatically to PostgreSQL hosted on Supabase in production. The curriculum is organised into seven thematic worlds and forty lessons. The first four worlds (thirteen lessons) teach sequencing, loops, conditions, variables, functions and debugging through drag-and-drop blocks on a grid, with a dual-representation engine that translates each block sequence into equivalent Python or JavaScript code in real time. The last three worlds (twenty-seven lessons) move learners into a text-based code editor where they write real Python, JavaScript and Java. Python runs as genuine CPython in the browser through Pyodide and WebAssembly, JavaScript runs inside an isolated evaluation scope, and Java is checked by a custom lexical analyser that simulates program output. Learners are motivated through experience points, daily coding streaks, twenty-six collectible badges, a timed Code Golf mode with a dynamic XP decay system, and a three-tier hint system
A COMPUTER-BASED FILL-IN-THE-GAP EXAMINATION SYSTEM
Manual preparation, invigilation, marking and compilation of examinations are slow, stressful and prone to error, particularly for large classes. Fill-in-the-gap examinations are harder to computerise than multiple-choice tests because students type their own answers, which must be compared with the acceptable responses while allowing for alternative spellings and minor typing errors. This report presents Fill-Gap, a computer-based fill-in-the-gap examination system developed by Group 3 for CPE 508. Fill-Gap is a three-layer web application with a React and TypeScript frontend, a NestJS REST API and a PostgreSQL database accessed through the Prisma ORM. Administrators manage faculties, departments, courses, lecturer-course assignments and student accounts, including bulk import from Excel. Lecturers create examinations, organise fill-in-the-gap questions by topic with multiple accepted answers and linked question chains, set how many questions are drawn from each topic, and register eligible students. When a student starts an examination, the server generates an individual paper by solving a chain-preserving subset-sum selection per topic and shuffling the result with the Fisher–Yates algorithm. Answers are autosaved, the deadline is enforced by the server, and expired attempts are submitted automatically. Answers are graded in Strict mode or in Non-Strict mode using Levenshtein edit distance, scores are preserved with a complete audit trail, and results for the whole class are exported to Excel.
Detecting BFRB Behavior using Sensor Data
This project presents a machine learning-based approach to detecting Body-Focused Repetitive Behaviors (BFRBs) such as hair pulling and skin picking using multimodal wearable sensor data. Addressing challenges like data imbalance, missing values in Time-of-Flight and Thermopile sensors, and noisy signals, the study utilized structured preprocessing, PCA-based feature reduction, and tree-based ensemble models. The experimental results, specifically using a Random Forest classifier, demonstrated high predictive performance with an F1-score of 0.88, confirming the feasibility of the proposed system for scalable, real-time behavioral health monitoring.
Fake or Real: The Imposter Hunt in Text
This project, part of the Fake or Real: The Imposter Hunt in Text Kaggle competition, focused on identifying authentic news articles from pairs consisting of one real and one synthetically altered version. Using transformer-based models like BERT, SBERT, and Longformer, the approach generated deep embeddings for the texts, which were then converted into interaction-based features such as vector differences and element-wise products. These features were used to train various classical machine learning classifiers. The Random Forest model achieved the highest performance with a 93.6% cross-validation accuracy and a leaderboard score of 0.87, demonstrating the effectiveness of embedding-based semantic comparison in detecting AI-generated misinformation.
Disease diagnosis using symptoms and patient data
This project developed a machine learning diagnostic system to predict diseases based on patient-reported symptoms, aiming to support healthcare workers in low-resource settings. By utilizing a unified dataset of over 32,000 patient records, the study addressed challenges such as overlapping symptoms and class imbalance. Using a tree-based modeling strategy, including models like Random Forest, XGBoost, and LightGBM combined into a soft voting ensemble, the system achieved high performance with a balanced F1-score across disease classes. The resulting tool, deployed via Streamlit, facilitates faster and more accurate triage decisions, demonstrating the practical application of data science in enhancing healthcare outcomes.
Design and Implementation of University Research Repository and Innovation Visibility System
The University Research Repository and Innovation Visibility System is a web-based platform developed to improve the organization and public discoverability of university research publications and patent records. The project was motivated by the need to bring research information that may otherwise exist across different personal, departmental, scholarly and intellectual property records into a structured platform. The system has two principal areas: a public discovery platform and a private researcher workspace. The public platform allows users to search and browse published publications, patents and researcher profiles without authentication. The private workspace allows authenticated university staff to manage their profiles and create, edit, publish and archive their publication and patent records. The system implements role-based access control with six roles: user, lecturer, department administrator, faculty administrator, IPTTO officer and super administrator. Authentication is custom-built using Argon2id password hashing, server-side sessions with hashed tokens and secure HTTP-only cookies. A major feature is automated publication import from OpenAlex. Researchers can search for and confirm their OpenAlex author identity, retrieve associated works, review the fetched candidates and promote selected works into draft publication records. Imported works are checked against existing records using DOI matching and, where a DOI is unavailable, normalized title and year matching. A staging model prevents external records from becoming publicly visible without researcher review and explicit publication. The system was implemented using React 19, TanStack Router, TanStack Start, TanStack Query, Vite 8, Tailwind CSS v4, Nitro, PostgreSQL and Drizzle ORM, and was deployed to production on Vercel. The completed system provides a focused foundation for university research visibility and researcher-managed research records.
Redesign Of The Department Of Computer Science And Engineering Masters In ICT Webpage
This project involved the redesign of the existing webpage for the Master’s degree program in Information and Communication Technology at the Department of Computer Science and Engineering. The objective was to modernize the webpage's appearance and integrate updated information for prospective students. The redesign was implemented using WordPress, drawing inspiration from contemporary academic website design trends.