Tools
Tools
Siddique, Jawed (2024) Deep Learning Framework for Land Use and Land Cover Classification and Change Detection. Masters thesis, Dublin, National College of Ireland.
Preview |
PDF (Master of Science)
Download (1MB) | Preview |
Preview |
PDF (Configuration Manual)
Download (3MB) | Preview |
Abstract
Land Use and Land Cover (LULC) classification and change detection have numerous environmental applications. However, the limited availability of high-quality labelled satellite data, high intra-class variability, and issues related to the noise in the available data LULC classification and change detection could be challenging. This study aims to address this challenge by performing LULC classification and detecting LULC change between the years 2018 and 2024 for the Greater Dublin Area by using the Sentinel-2 satellite imagery dataset and Deep Learning architecture. This study used data pre-processing and data transformation techniques to prepare a multi-spectral tiled satellite image dataset before employing supervised classification techniques. For classification purposes, a deep, multilayered CNN architecture ResNet50 model was used and five major classes: Artificial Surfaces, Agricultural Areas, Forest and Seminatural Areas, Wetlands, and Water Bodies were identified. For LULC classification evaluation, the classification error matrix, accuracy, precision, recall, F-Score and kappa analysis were used. For LULC change detection, the differences in the areas covered by each land cover class between the periods are measured. According to the findings from the study, the Agricultural land cover class accounted for 74.67% in 2018 and 75.27% in 2024, making it the most extensive class. There is, however, a shift in the second-largest class cover from Forest and Seminatural Areas in 2018 to Artificial Areas in 2024. Additionally, this study identified a significant change in the Forest (-3.20%) and Artificial Surfaces (+1.24) from 2018 to 2024. Lastly, the classification methodology achieved overall accuracy, precision and Kappa statistics of 92.38%, 92.41, and 0.91 respectively.
| Item Type: | Thesis (Masters) |
|---|---|
| Supervisors: | Name Email Stynes, Paul UNSPECIFIED Jilani, Musfira UNSPECIFIED Cudden, Mark UNSPECIFIED |
| Uncontrolled Keywords: | Land Use; Land Cover; Change Detection; Sentinal-2; Deep Learning; Convolutional Neural Networks; ResNet50; Accuracy Assessment |
| Subjects: | H Social Sciences > HD Industries. Land use. Labor Q Science > QA Mathematics > Electronic computers. Computer science T Technology > T Technology (General) > Information Technology > Electronic computers. Computer science Q Science > Q Science (General) > Self-organizing systems. Conscious automata > Machine learning |
| Divisions: | School of Computing > Master of Science in Data Analytics |
| Depositing User: | Ciara O'Brien |
| Date Deposited: | 26 Aug 2025 10:39 |
| Last Modified: | 26 Aug 2025 10:39 |
| URI: | https://norma.ncirl.ie/id/eprint/8632 |
Actions (login required)
 |
View Item |