A Edge-Guided Satellite Image Semantic Segmentation Method for Real Estate Appraisal

This paper proposes an edge-guided satellite image semantic segmentation method for real estate appraisal. The digitization of real estate appraisal has been significantly enhanced by advancements in computer vision technologies, enabling automated recognition and analysis of property features from satellite images, aerial photographs, and street view images. However, current methods face limitations due to the resolution and quality of satellite imagery, which affect the precision of evaluations. To address these challenges, this study integrates semantic segmentation with a mathematical house pricing valuation model. The Edge Graph Attention U-Net is introduced, focusing on the edges and details of property features to enhance seg- mentation precision. The proposed semantic valuation algorithm integrates segmentation results with traditional market data and Geographic Information System data, employing mathematical model for property value prediction. Experimental results demon- strate the efficacy and accuracy of the proposed model across multiple cities and property types, providing a robust solution for modern real estate appraisal. The primary contributions include improved edge detection for segmentation, integration of detailed segmentation with valuation algorithms, and validation of the model's effectiveness in realworld applications.