Ashesi Institutional Repository
Welcome to the Ashesi Institutional Repository (AIR), an archive for preserving and sharing scholarly work at Ashesi University.
Contributors to the repository ensure that their scholarly and creative work is preserved, indexed and showcased for a global audience. Students who produce strong research work have the privilege of getting their work published on AIR. Ashesi Community members can make submissions through warrenlibrary@ashesi.edu.gh.
The repository is organized in "Communities" that group publications by department and/or subject. You can browse the "Community" by Collection, Author, Title, or Issue Date. Alternatively, a search function enables easy access to information.
Communities
Select a community to browse its collections.
- Faculty and student publications on arts, social sciences and humanities subjects
- Faculty and student publications on STEM subjects
- An archive of all university publications
Recent Submissions
Improving public transportation for people with disabilities in Ghana through digital technologies
(2024-08) Wood, Henry Michael
Despite being valued members of society, in Ghana, people with disabilities (PWDs) are often overlooked when it comes to public transportation, making it difficult for them to perform their day-to-day activities. Public transportation services in Ghana need more fundamental elements that will allow PWDs to use their services more seamlessly, discouraging them from using the more expensive private transportation services that continue to promote traffic. This research aims to look at the current state of public transportation for PWDs, the measures being put in place by the government to curb these issues, how digitalization is changing the public transportation space, and how this digital transformation can be used to benefit PWDs. This paper will look to develop a practical solution to some of the problems PWDs face by creating a web application programming interface (API) that can be used as a companion for existing transportation applications in Ghana.
CODET: An artificial intelligence-based multisensory system for precision harvesting and purchase decisions
(2024-08) Awariyah, Edwin Nsoh
Fruit ripeness detection plays a key role in precision agriculture, which enables optimal harvesting and transportation of produce to retailers. Various methods have been explored as means to detect this feature; however, in modern age, no standardised method exists to classify fruits based on their ripening state. Thus, this project aims to design and develop a smart system based on Artificial Intelligence to evaluate the ripeness state. The result is a portable handheld device capable of detecting maturity based on the physical properties of the fruit in real-time.
Codet is a portable fruit ripeness prediction system with hardware and software components that make these predictions possible. It consists of an RGBC sensor, an 18-channel spectrometer (from 410nm to 940nm), an 11-channel spectrometer (from 430nm to 670nm), an ultrasonic sensor, time of flight sensors and a camera. With the help of a display, the user can determine the ripeness state of a fruit with just a button press. The device can identify the ripeness state of fruits with an accuracy of 99.9%.
DoverColl: A system that makes use of IoT to enhance efficiency in waste collection of urban households in ghana
(2024-08) Opoku-Boadu, Bernd Osafo
Traditionally, if an urban household wants to dispose of the waste generated, they subscribe to a company's collection service such as Zoomlion which is periodic. However, if there is waste to be collected before the collection date, it leads to an overflow of waste which has potential health risks and brings about foul odours. DoverColl is a waste management system that stream lines the waste collection process by connecting households with waste collectors in real-time with an IoT component. Waste collectors are able to find households easily as a result of routing applications like Google Maps. The system employs a hybrid methodology of the reuse-oriented and agile methodology. This hybrid methodology chosen for this project is suitable because some components of the project exist already, so building one from scratch is less cost-effective and might not be as efficient. These components have been used in working systems, which would increase the system's availability and dependability. The project is also relatively large, and there is a need to make room for changing requirements to ensure that the solution we are building is relevant to the stakeholders and is a sustainable solution.
Enhancing equitable resource allocation: Integrating banker's algorithm with machine learning prediction
(2024-08) Tetteh, Abigail Efua
Global organizations play a pivotal role in tackling some of the world's socio-economic challenges. With well-defined processes, organizations such as UNICEF (United Nations International Children Emergency Fund) make well-informed decisions pertaining to their resource allocation approaches. Although these approaches have recorded some level of efficiency by ultimately allocating resources, they are heavily centered on traditional approaches that fail to reflect the dynamic changes to the growing needs of society due to their static nature. Just like the Public Finance toolkit used by UNICEF, such approaches rely on a paper-like streamlined, human-generated index that determines how allocation works, and this erases the overarching goal of resource allocation, fairness. This capstone focuses on optimizing resource allocation within organizations like UNICEF through the integration of various scheduling algorithms with Banker's algorithm. The goal is to streamline the process of resource allocation right from handling funding requests from the public and subsequently distributing available resources based on pre-defined conditions of the algorithm. The web application relies on PHP-implemented algorithms to manage requests and perform resource allocation to achieve more efficient and transparent results.
The system allows organizations to input total available resources and select from multiple scheduling algorithms, including First Come First Serve (FCFS), Priority-Based and Round Robin. The Banker's algorithm is employed to determine the feasibility of each funding request, dynamically updating the request status to either "Accepted" or "Rejected" based on the allocation results. The backend integrates robust database operations to store, update, and retrieve request statuses and allocation details, ensuring data integrity and consistency. User authentication and session management are implemented to secure the application and personalize user experience.
This project aims to provide a scalable and efficient solution for resource management in organizations, facilitating better decision-making and enhancing operational efficiency. The system's design and implementation address key challenges in resource allocation, offering a comprehensive tool for managing and optimizing funding requests.
Vision Vantage: A decision support system for footbal player scouts
(2024-08) Nettey, Sandra
The rapidly evolving sports technology landscape, characterized by advancements in data analytics, machine learning and real-time processing, has paved the way for innovative solutions in player and game analysis. This paper presents Vision Vantage, a cutting-edge football analysis web application harnessing computer vision and machine learning to overcome the limitations of traditional player evaluation methods. Vision Vantage integrates video processing with comprehensive statistical analysis to offer a transformative approach to game assessment and player performance evaluation. The core functionalities of Vision Vantage include object detection using YOLOv5, multi-object tracking with Byte Track, and backend management through FastAPI. These technologies collectively enhance the accuracy of player and ball tracking, enable speed and distance estimation, and facilitate pass detection.
The system's architecture supports seamless user registration, game creation, video uploads and skill management, ensuring a robust and user-friendly experience. Rigorous testing phasis, including unit and component testing, validated the system's functionality and performance. Unit tests confirmed the effective handling of valid and invalid inputs across various endpoints, while component testing, conducted using Cypress, verified the smooth operation of frontend interfaces. User testing further refined the system based on feedback, leading to a transition from Django to FastAPI for improved machine learning model integration. Despite its successes, Vision Vantage faces limitations, notably inconsistencies in tracker IDs and challenges related to processing large video files. These issues impact data integrity and system performance, highlighting areas for future improvement. Recommendations for future work include the development of a player highlight component to dynamically showcase high-performing players based on statistical analysis.
This enhancement aims to provide deeper insights into player contributions and improve the scouting experience by offering value performance metrics and visual indicators. Vision Vantage represents a significant advancement in football analysis technology, provide a comprehensive tool for player evaluation and game assessment.