Detecting Cadastral Boundary from Satellite Images Using U-Net model

2025-02-16

AI Summary

To address the challenges of low accuracy and scarce annotated data in automatic extraction of farmland cadastral boundaries, this paper proposes an improved U-Net–based semantic segmentation method. First, ResNet34 is integrated as the backbone network into the U-Net architecture, and transfer learning is employed to adapt the model to small-sample remote sensing imagery of farmland. Second, a three-class pixel-wise classification scheme—boundary, field parcel, and background—is adopted to explicitly encode geometric characteristics of cadastral boundaries. Experiments conducted on multi-temporal Sentinel-2 satellite imagery over Iran demonstrate that the method achieves 88% precision, 75% recall, and an F1-score of 81%. Compared with conventional approaches, it significantly improves boundary localization accuracy and generalization capability. The proposed framework provides a transferable technical solution for high-precision cadastral updating in resource-constrained regions.

Technology Category

• Computer Vision: Remote Sensing / Geospatial AI
• Machine Learning: Multi-class/Multi-label Learning & Extreme Classification
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Web crawling and indexing
• Web Mining and Content Analysis: Robustness and generalizability of Web mining methods

Abstract

Finding the cadastral boundaries of farmlands is a crucial concern for land administration. Therefore, using deep learning methods to expedite and simplify the extraction of cadastral boundaries from satellite and unmanned aerial vehicle (UAV) images is critical. In this paper, we employ transfer learning to train a U-Net model with a ResNet34 backbone to detect cadastral boundaries through three-class semantic segmentation:"boundary","field", and"background". We evaluate the performance on two satellite images from farmlands in Iran using"precision","recall", and"F-score", achieving high values of 88%, 75%, and 81%, respectively, which indicate promising results.

Problem

• Detecting farmland cadastral boundaries
• Using U-Net model with ResNet34 backbone
• Three-class semantic segmentation from satellite images

Innovation

• Uses U-Net with ResNet34 backbone
• Applies transfer learning for training
• Performs three-class semantic segmentation