Rashmi Sonth

Publication Date

Spring 2025

Degree Type

Master's Project

Degree Name

Master of Science in Computer Science (MSCS)

Department

Computer Science

First Advisor

Navrati Saxena

Second Advisor

Abhishek Roy

Third Advisor

Robert Chun

Keywords

Semantic segmentation, land use classification, deep learning, U-Net, multi-spectral imagery, urban planning, TOPSIS.

Abstract

Accurate land use classification is the backbone for urban planning. But with poor quality satellite images, varied landscapes and structures which are changing faster than ever, it becomes a challenge to define clear boundaries and hence to urban planning. This research explores the application of deep-learning model for land use classification and asses the suitability of the land. The proposed model combines a multi-scale U-Net architecture with Transformer blocks applied on a multi-spectral satellite images that improves the semantic segmentation greatly across the urban and rural regions. Additionally, a patch-wise segmentation is applied to overcome the common problem of feature imbalance and boundary ambiguity. The work extends beyond by determining the suitability scores of the regions using the Technique for Order Preference by Similarity to Ideal Solution (TOPSIS) that indulges in the adaptive feature weighting. The framework has been tested on real-world datasets to evaluate the performance in classifying the land cover, that can result in actionable insights. The proposed approach when compared against existing methodologies shows significant improvements with enhanced segmentation accuracy, sharper boundary definitions, and clearer suitability mapping, achieving gains of up to 172.94% in certain land use classes. By integrating advanced deep learning techniques with geospatial decision-making models, the proposed approach offers a scalable solution for sustainable urban development.

Download

Available for download on Sunday, May 10, 2026

Share

COinS