Institutional-Repository, University of Moratuwa
Welcome to the University of Moratuwa Digital Repository, which houses postgraduate theses and dissertations, research articles presented at conferences by faculties and departments, university-published journal articles and research publications authored by academic staff. This online repository stores, preserves and distributes the University's scholarly work. This service allows University members to share their research with a larger audience.
![]() Research Publications | ![]() Thesis & Dissertation | ![]() E- Books |



![]() UoM Journal Publications | ![]() UoM Conference Proceedings | ![]() Articles published in Scimago's Q1 journals | ![]() UoM Research Reports | ![]() Other Articles authored by UoM staff |
Recent Submissions
item: Thesis-Abstract
Sustained context-aware image generation for the Dungeons & Dragons domain
(2024) Weerasundara, WAG; De Silva, N
Dungeons & Dragons (D&D) is a fantasy tabletop role-playing game which has become a subculture phenomenon due to its immense popularity. In a D&D adventure images play a major role in guiding the players’ emotions and providing additional information regarding the setting. The proposed research attempts to use text extracted from premade adventure to generate cohesive and contextually accurate images according to the given setting. The core of the proposed methodology involves two strategic components. First, the project seeks to develop an Natural Language Processing (NLP) model capable of deep textual analysis to identify and understand the underlying context and key elements within the D&D text. This model will focus on extracting salient features and narrative cues from the text, which will then be used to generate precise prompts. These prompts are designed to encapsulate the essential elements needed to guide the image generation process, ensuring that the resulting visuals are not only relevant but also enrich the storytelling by aligning closely with the D&D lore. The second component is a comprehensive pipeline that generates consistent images in a zero-shot manner. Additionally, this study proposes the creation of an end-to-end pipeline that not only generates images but also automates the creation of complete D&D adventures. This pipeline will integrate multi-agent workflows, an image generation framework, and NLP techniques to produce a comprehensive suite of adventure materials—including story narratives, gameplay guidelines, gameplay-related tables, contextual images, and maps. This holistic approach is designed to streamline the preparation process for Dungeon Masters, enabling them to deliver richly detailed sessions with less preparation time.
item: Thesis-Abstract
Energy performance benchmark analysis of supermarket buildings in Sri Lanka
(2024) Sumathiratna, WCC; Wijewardena, A; Suraweera, D
In the recent past, the energy consumption in Sri Lanka has been drastically increased due to the rapid urbanization, improved living standards and the change of consumption patterns of the people, resulting in a long-term sustainability issue for the people. A report published by Sustainable Energy Authority (SEA) of Sri Lanka has indicated that, in 2020, 67.6% of electricity was consumed by the domestic and commercial buildings. Because they utilize air conditioning and refrigeration equipment more frequently than other commercial buildings, retail facilities like supermarkets also require a lot of energy. Each and every other store in Sri Lanka is dependent on the grid power supply, even though relatively few supermarket companies produce electricity using solar photovoltaic systems.. Increasing the number of supermarkets in par with the demand, increasing the load on the national grid also escalates. So far, Sri Lanka has developed and published the energy benchmarks for commercial sector, hospitality sector, tea processing sector and the apparel sector. Therefore, it is crucial to quantify and establish the energy benchmark for retail supermarket sector in Sri Lanka to understand and regulate the energy performance of those buildings. In this study the electrical energy consumption was the only energy type considered as other energy sources like LP gas and gasoline use were negligible and past data for five years were unable to access. The average specific energy usage, or average Energy Use Intensity (EUI), of the Sri Lankan supermarkets assessed was discovered to be 817 kW/m2.year for supermarkets with air-conditioned sales spaces and walk-in cold rooms (heavy electrical energy usage) and 244 kW/m2.year for government-owned stores with the least amount of air conditioning (nominal use of electrical energy). Hence energy benchmarking of the retail supermarkets is crucial as it is complex to understand their energy performance with comparative to other type of buildings. Moreover, during this study the energy consumption breakdown for Sri Lankan supermarkets with walk in cool rooms and air conditioned sales area was calculated. The energy consumption percentages are, refrigeration 48%, space cooling 19%, ventilation 3%, exterior lighting 4%, miscellaneous use 23% including electrical appliances in kitchen and bakeries.
item: Conference-Full-text
International conference on Facilities Management Futures (FMF) 2025: Safety-Enabled and Sustainable Facilities
(Facilities Management Research Unit (FaMRU), 2025) De Silva, N; Sridarran, P
item: Conference-Full-text
Developing evaluation criteria for assessing the performance of smart retrofitting of existing buildings
(Facilities Management Research Unit (FaMRU), 2025) Alahakoon, DOA; Mapa, MMIS; Silva, SDRY; De Silva, N; Sridarran, P
The existing building sector is contributing to high energy consumption and CO2 emissions worldwide. Therefore, building practitioners should commit to adopting energy-conserving practices and minimizing CO2 emissions. Consequently, the adoption of smart retrofitting (SR) in existing buildings is one of the easiest and cheapest ways to achieve energy efficiency and reduce CO2 emissions. In Sri Lanka, SR adaptation was at a low level. However, there is a shortage of research and studies concerning assessing the performance of the SR of existing buildings in Sri Lanka after its implementation. Therefore, the purpose of this research is to develop an assessment model to assess the performance of the SR of existing buildings for the post-installation phase in Sri Lanka. A qualitative research method was used in this research to identify factors to develop an SR performance assessment model for existing buildings. Semi-structured expert interviews were conducted to validate the identified factors and obtain additional information related to the SR assessment, and the data collected was analysed through manual content analysis. As a result, a total of 39 factors were identified, which fell into seven categories: technical adaptability, human comfort, economic adaptability, energy management, environmental sustainability, safety and security, and compliance. Finally, the SR performance assessment model is developed.
item: Thesis-Abstract
Sri Lankan elephant sound classification using deep learning
(2024) Dewmini, AGHUD; Meedeniya, D
Understanding elephant caller types is crucial for various aspects of wildlife conservation and ecological research. By decoding the intricate vocalizations of elephants, researchers gain valuable insights into their behavior, social dynamics, and emotional expressions, which are pivotal for species conservation efforts. Elephant vocalizations serve as indicators of ecosystem health and vitality, aiding in ecological monitoring and biodiversity conservation initiatives. Furthermore, investigating caller types contributes to the preservation of cultural heritage by honoring the profound connection between humans and elephants across generations. In essence, delving into the world of elephant communication not only advances scientific knowledge but also fosters harmony between humans and these majestic animals, ensuring their long-term survival in the wild. In this study, we delve into the domain of elephant caller-type classification utilizing raw audio format processing. Our focus lies on exploring lightweight models suitable for deployment on edge devices, including MobileNet, YAMNET, and RawNet, alongside introducing a novel model termed ElephantCallerNet, based on ACDnet architecture. Notably, our investigation reveals that the ACDnet-based ElephantCaller- Net achieves an impressive accuracy of 89% when applied to a raw audio dataset. Leveraging Bayesian optimization techniques, we fine-tune crucial parameters such as learning rate, dropout, and kernel size, thereby enhancing model performance. Moreover, we scrutinize the efficacy of spectrogram-based training, a prevalent approach in animal sound classification. Through comparative analysis, we ascertain that for our dataset, raw audio processing outperforms spectrogram-based methods. In contrast to other models in the literature that primarily focus on a single caller type or binary classification (such as identifying whether a sound is an elephant voice or not), our models are designed to classify three distinct caller types: Roar, Rumble, and Trumpet. This approach significantly increases the complexity of our experiments compared to those discussed in the literature. In the domain of elephant vocalization analysis, there has been limited exploration into the direct processing of raw audio data. Predominantly, various feature extraction techniques have been employed before training machine learning algorithms. In our investigation, we aim to bypass preprocessing stages and directly input raw audio data into machine learning models to assess the feasibility and efficacy of training on unprocessed audio signals..