LTAT bakalaureusetööd – Bachelor's theses

Selle kollektsiooni püsiv URIhttps://hdl.handle.net/10062/32748

Sirvi

Nüüd näidatakse 1 - 20 1525

Meiobentose hulka kuuluvate organismide automaatne tuvastamine masinõppe meetoditega
(Tartu Ülikool, 2025) Lokko, Külli; Jaanuska, Ljubov, juhendaja; Tartu Ülikool. Loodus- ja täppisteaduste valdkond; Tartu Ülikool. Arvutiteaduse instituut
The objective of the bachelor's thesis was to automate the detection and identification of meiobenthos in micrographs using machine learning methods. Meiobenthos consists of many different groups of microscopic organisms living in the aquatic sediments. In aquatic environments, meiobenthos plays an important role in energy flow and nutrient cycles. Counting and identifying these organisms manually is very labor-intensive and time-consuming, and requires extensive training. Therefore, there is a clear need for automation, which would greatly facilitate abundance and biomass estimation, taxonomic composition assessment, and the study of the ecological relationships and nutrient cycles. First, an annotated database of micrographs was created as part of the thesis. To ensure high-quality training data, each meiobenthic organism in the images was manually annotated. Prior to this work, no comparable dataset was publicly available. Next, models were trained using two object detection models (Faster R-CNN and YOLO11) to automatically detect all meiobenthic organisms in the images and to identify their taxonomic group. The detection performance of the best models was comparable to that of existing models trained on plankton datasets, achieving an error rate of only 3.7% in abundance estimation on the test set. The models performed slightly less well in identifying the correct organism group (taxon). The best Faster R-CNN model correctly identified both the location and taxonomic group of organisms in 86.3% of cases while the YOLO11 model identified both correctly in 80.7% of cases.
Õppimist toetav veebirakendus kitarrimängijatele
(Tartu Ülikool, 2025) Meos, Rasmus; Aller, Sven, juhendaja; Tartu Ülikool. Loodus- ja täppisteaduste valdkond; Tartu Ülikool. Arvutiteaduse instituut
The aim of this Bachelor’s thesis was to develop a web application that helps guitar players find songs that match their skill level. The user can enter the name of an artist, after which the system displays a list of songs along with a suitability percentage—calculated based on the chords the user has previously marked as learned. The thesis provides an overview of key music theory concepts, including chord structure, transposition, and simplification. It also discusses web scraping techniques used to collect chord sheets, and introduces the technologies and system architecture selected for implementing the application. The thesis further describes the custom-built chord processing algorithms, including key detection, chord transposition, and simplification logic. The final chapters present measurement results, user feedback, and potential opportunities for further development.
PCB komponentide tuvastamine
(Tartu Ülikool, 2025) Tiitson, Joonas; Raun, Kristo, juhendaja; Tartu Ülikool. Loodus- ja täppisteaduste valdkond; Tartu Ülikool. Arvutiteaduse instituut
The mission of this paper is to create machine learning based models for detecting the orientation of printed circuit board (PCB) components. To achieve the desired goal, during the course of this work it was necessary to create a photo-bank of different kinds of annotated PCB-s, two models, one based on Faster-RCNN with a ResNeXt backbone and the other based on YOLO. A few validation and testing programs were created to compare the upsides and downsides of each model. During the development process, it was a general rule to try use open-source solutions, so that the created programs and models would be free to use for anyone who is interested. The completed project allows anyone to connect their annotated project to model training files, to use them for directional object detection.
Designing Textual Representations of Business Process Model and Notation for Large Language Models: A Research in Prompting Strategies and Automated Evaluation
(Tartu Ülikool, 2025) Pärloja, Karl Mattias; Chapela De La Campa, David, juhendaja; Tartu Ülikool. Loodus- ja täppisteaduste valdkond; Tartu Ülikool. Arvutiteaduse instituut
Business Process Model and Notation (BPMN) is a widely used standard for modelling and analysing business processes. However, its graphical nature and the complexity of its underlying XML representation pose challenges for seamless integration with Large Language Models (LLMs), which primarily process textual data. The objective of this thesis is twofold: to explore the design and development of optimised textual representations by systematically converting BPMN models into structured textual formats to enhance LLM comprehension and to design a benchmark to automatically evaluate the effectiveness of different prompt strategies in enabling LLMs to interpret BPMN structures.
Enhancing the Virtual Reality Experience Psyrreal
(Tartu Ülikool, 2025) Käär, Erko; Vasser, Madis , juhendaja; Tartu Ülikool. Loodus- ja täppisteaduste valdkond; Tartu Ülikool. Arvutiteaduse instituut
This bachelor’s thesis focuses on enhancing the user interface of the psychedelic virtual reality application Psyrreal to improve its usability in therapeutic contexts. A new intuitive menu system and visual effects were integrated based on feedback from Psyrreal’s beta testing. Customizable scenes, developed using Unreal Engine 5.3.2, support the psychedelic experience. Usability testing confirmed that the improved interface is intuitive for novices and aligns with therapeutic goals.
Tootmisliini kvaliteedijälgimise süsteemi andmebaasi ning töölaua loomine
(Tartu Ülikool, 2025) Kallaste, Harmo-Ron; Tampuu, Ardi, juhendaja; Tartu Ülikool. Loodus- ja täppisteaduste valdkond; Tartu Ülikool. Arvutiteaduse instituut
Quality control is an important step in each production chain. Visual verification of the results is needed to understand the quality and compliance with requirements of the production. This thesis describes how a dashboard, database, and a backend system were developed for a production line quality control system in Nõo Lihatööstus, helping the client get an overview of the quality and the common issues of their packaging in real-time and also through historical data.
Koodianalüsaatori arendus kursuse „Programmeerimise alused“ projektide tehnilise analüüsi automatiseerimiseks
(Tartu Ülikool, 2025) Pihlak, Merili; Suviste, Reelika, juhendaja; Tartu Ülikool. Loodus- ja täppisteaduste valdkond; Tartu Ülikool. Arvutiteaduse instituut
Evaluating and performing analysis to over 100 projects in a programming course is a timeconsuming task. With this thesis, a code analyzer was proposed and developed for the course “Introduction to Programming” to automatically analyze projects. The result was a code analyzer, which in 2 minutes, outputted code analyses of over a hundred projects into a single summarized Excel file, where each project has 48-50 different technical analysis metrics.
Automaattestide kirjapanemise keele TSL edasiarendamine
(Tartu Ülikool, 2025) Saviauk, Karmo; Palm, Reimo, juhendaja; Tartu Ülikool. Loodus- ja täppisteaduste valdkond; Tartu Ülikool. Arvutiteaduse instituut
Automated assessment plays a key role in programming education by providing immediate feedback to students and reducing the workload of instructors. This thesis focuses on the development of TSL (Test Specific Language), a domain-specific language used at the University of Tartu for writing automated tests. The work includes an analysis of the most common types of programming assignments in the “Computer Programming” course and the corresponding assessment strategies. To improve the TSL language, new test types were implemented, existing ones were enhanced, and the quality of feedback provided to students was improved. All changes were integrated into the TSL language to expand its capabilities and improve its practical applicability. As a result of this work, the system became more capable of detecting common errors, evaluating more complex solutions, providing clearer and more meaningful feedback to students, and reducing the workload of instructors.
Andmeturbe kursuse praktikumimaterjalide värskendamine ja täiendamine
(Tartu Ülikool, 2025) Vaiküll, Hjalmar; Peets, Alo, juhendaja; Tartu Ülikool. Loodus- ja täppisteaduste valdkond; Tartu Ülikool. Arvutiteaduse instituut
Due to the rapidly evolving cyber landscape, it is essential to ensure the relevance and effectiveness of educational materials related to cybersecurity. As part of this bachelor’s thesis, the lab materials for the University of Tartu course LTAT.06.002 “Computer Security”, which is taken by over 300 students annually, were updated and enhanced. The goal of the thesis was to test and modernize all 15 lab guides and associated resources, such as virtual machines, web applications, practical tests and the grading guide. Key achievements of the thesis include the update of 4 preconfigured virtual machines, the creation of 3 Moodle quizzes with automatically graded practical tasks, modernization of various software tools and their configuration files, and a comprehensive overhaul of the content and layout of all lab guides. The thesis provides a detailed description of all activities related to the practicals and explains the reasoning behind the updates and improvements made.
Developing a Mesh Networked IoT System for Plant Health Monitoring
(Tartu Ülikool, 2025) Raaga, Toomas Hendrik; Norbisrath, Ulrich, juhendaja; Zorec, Matevž Borjan, juhendaja; Tartu Ülikool. Loodus- ja täppisteaduste valdkond; Tartu Ülikool. Arvutiteaduse instituut
This thesis presents Smart Garden, a mesh-networked IoT system that addresses the challenge of determining the right time to water indoor plants. Unlike existing solutions that operate as standalone units, Smart Garden employs sensors in a mesh network, enabling monitoring in areas with poor coverage. The system continuously tracks soil moisture via capacitive sensors, transmits data through a mesh network, and notifies users through a web interface when plants need watering. Testing demonstrated 96% reliability over three months and successful communication through concrete barriers. Evaluation confirmed that the system effectively transforms the question of ”When do I water my plants?” into actionable notifications based on actual soil moisture data.
Veebipoe loomine brändile Meemehed
(Tartu Ülikool, 2025) Tann, Morris; Janno, Jaan, juhendaja; Tartu Ülikool. Loodus- ja täppisteaduste valdkond; Tartu Ülikool. Arvutiteaduse instituut
In this bachelor's thesis, a web store was created for the brand Meemehed, which produces and sells flavored honey. The company previously lacked a digital sales channel, limiting product visibility and the growth of its customer base. The aim of the project was to develop a web application that enables product browsing, shopping cart management, secure payments, and user account creation. Special attention was given to performance, security, and user experience to ensure a competitive shopping environment. The developed web store supports the company’s digital development, helps improve its market position and reach new customers.
Enhancing Single Packet Authorization with Multi-Factor Authentication
(Tartu Ülikool, 2025) Lepp, Rico-Andreas; Oja, Tarmo, juhendaja; Tartu Ülikool. Loodus- ja täppisteaduste valdkond; Tartu Ülikool. Arvutiteaduse instituut
Single Packet Authorization (SPA) is a network security technique for adding an additional layer of authentication and concealing services from adversaries while still allowing access for authorized users. This shifts the attack surface from the services to the SPA implementation itself, making its security crucial. Publicly available implementations of SPA use a shared static secret, leaving a single point of failure in case the secret is compromised. One possible solution to this issue is the usage of Multi-Factor Authentication (MFA) with the Time-based One-Time Password (TOTP) system, which uses short-lived codes generated on an external device. As a result of this thesis, the TOTP system was integrated with a specific implementation of SPA known as fwknop.
Implementing and Testing a Simple Algorithm for Consistent Query Answering
(Tartu Ülikool, 2025) Kimst, Mattias; Hannula, Miika Juhani, juhendaja; Tartu Ülikool. Loodus- ja täppisteaduste valdkond; Tartu Ülikool. Arvutiteaduse instituut
To address inconsistencies in databases, data cleaning is commonly used; however, it can be complex or may result in the loss of some data. As an alternative, the consistent query answering (CQA) paradigm has been developed to provide consistent query results without altering the database. In this thesis, an algorithm for CQA for primary key violations and conjunctive queries is implemented in the Java programming language and its performance is evaluated on synthetically generated data. Additionally, the database purification technique is tested as an option to enhance the algorithm’s performance. The results indicate that purification significantly improves performance. However, regardless of the use of purification, the algorithm’s runtime increases rapidly as the database size grows.
Konteineripõhise keskkonna loomine TÜ kliimateenuste portaali jaoks
(Tartu Ülikool, 2025) Nigul, Jaan; Luhamaa, Andres, juhendaja; Tartu Ülikool. Loodus- ja täppisteaduste valdkond; Tartu Ülikool. Arvutiteaduse instituut
The aim of this bachelor’s thesis was to develop a container-based platform for sharing and visualizing various climate datasets, enabling the creation of configurable and interactive dashboards suitable for both specific applications and rapid prototyping. As a result, a scalable system was built that allows launching personalized dashboards, managing users and sessions, and utilizing data stored both locally and in the cloud. The system includes user authentication, container and session management, logging and monitoring functionalities, and is built on Amazon Web Services and Panel framework.
Kasutajakogemuse disain andmete visualiseerimisel paketi TrajectoryViz disainiuuenduse näitel
(Tartu Ülikool, 2025) Sooman, Hanna Mia; Pajusalu, Maarja, juhendaja; Tartu Ülikool. Loodus- ja täppisteaduste valdkond; Tartu Ülikool. Arvutiteaduse instituut
This thesis focuses on improving the user experience design of the TrajectoryViz package by identifying existing issues and proposing potential redesign, taking into account the specific characteristics of data visualization tools. As part of the research, one iteration of the design thinking framework design process was conducted, during which a prototype was developed and tested with users. Based on the results, suggestions are made for further development of the tool.
Mängutuur – visuaalne arenduskeskkond õppemängu jaoks
(Tartu Ülikool, 2025) Kulu, Mihkel; Nael, Daniel, juhendaja; Tartu Ülikool. Loodus- ja täppisteaduste valdkond; Tartu Ülikool. Arvutiteaduse instituut
This thesis presents the design and implementation of “Game Tour”, a development environment created to support the further development of the educational game “Virtual Tour Game”. The environment allows for visual management of in-game elements such as navigation buttons, video screens, and characters. The study analyzes similar games and tours that influenced the structure of “Virtual Tour Game” and explores the potential of educational games in learning contexts. The core functionality and technical architecture of the developed environment are described. Based on user testing, the environment is visually clear and functional, but requires further refinement before being suitable for broader use.
Syntax Bee: An Interactive Prototype for Investigating the Impact of Code Visualisation on the Motivation in Learning Programming
(Tartu Ülikool, 2025) Ambur, Brigitta; Spors, Velvet, juhendaja; Tartu Ülikool. Loodus- ja täppisteaduste valdkond; Tartu Ülikool. Arvutiteaduse instituut
The purpose of this thesis was to create an interactive prototype to investigate how code visualisation affects students’ intrinsic motivation while learning programming. The theoretical part discusses learning to program, motivation in learning to program, visualisation in education, and code visualisation. The practical part consists of interviews and a quasi-experiment. The interviews focused on participants’ experience with learning programming. The experiment involved creating an interactive prototype based on the interview results, testing the prototype, and analysing the results. Motivation was measured before and after testing the prototype by using adapted version of the Intrinsic Motivation Inventory. The results showed that intrinsic motivation increased among participants who used the visualised prototype, whereas motivation decreased in the control group.
Search for Good Quantum Error-Correcting Codes
(Tartu Ülikool, 2025) Aitsen, Daglas; Bocharova, Irina, juhendaja; Tartu Ülikool. Loodus- ja täppisteaduste valdkond; Tartu Ülikool. Arvutiteaduse instituut
Quantum error correction codes are developed to correct errors that occur in quantum data. However, their design is complicated by the fundamental constraints of quantum mechanics. This thesis explores how to adapt quasi-cyclic low-density parity-check (QC-LDPC) codes, used in classical communications, for quantum error correction. Since a small girth in an LDPC code is linked to poor decoding performance, this thesis focuses on constructing quantum QC-LDPC codes with large girth. To achieve this, we present two methods: one based on minimal-weight codewords and another based on multiplicity constraints to avoid short cycles. Using these approaches, we construct quantum QC-LDPC codes with girth 8. The construction methods offer a more flexible alternative to existing techniques, enabling the design of large-girth codes with a wider range of structures and sizes.
Kahendpuu- ja kuhjaalgoritmide läbimängija ja hindaja
(Tartu Ülikool, 2025) Michelis, Markus; Põder, Ahti, juhendaja; Tartu Ülikool. Loodus- ja täppisteaduste valdkond; Tartu Ülikool. Arvutiteaduse instituut
The aim of this bachelor's thesis was to develop an interactive application for the University of Tartu course “Algorithms and Data Structures” to simulate the step-by-step execution of binary search tree, AVL tree, and heap algorithms. The developed program allows users to interactively work through these algorithms while automatically logging all mistakes made during the process. This provides a foundation for objective and efficient assessment, replacing the currently time-consuming process of preparing and grading written tests. The application is designed to be integrated into the DeepMOOC e-learning platform, offering a solution for both learning and evaluating algorithmic understanding.
Eestikeele sisukokkuvõtja edasiarendamine
(Tartu Ülikool, 2025) Peedosk, Marko; Aller, Sven, juhendaja; Tartu Ülikool. Loodus- ja täppisteaduste valdkond; Tartu Ülikool. Arvutiteaduse instituut
In an information-rich environment, automatic summarization helps save time by highlighting key content and enabling faster text navigation. EstSum was enhanced with semantic analysis: two linguistic modules were added, and transfer learning was tested using a few-shot approach to evaluate performance with limited training data. The wordnet-based module slightly improved accuracy but significantly slowed the system, while the language model–based module achieved a negligible increase in accuracy without affecting speed. Although few-shot performance matched the baseline, further improvements are possible by supplying more training examples and refining prompts. To improve accessibility, a web-based user interface was also developed for non-technical users.

Sirvi

Viimati lisatud