Building Detection in High-Resolution Remote Sensing Images by Enhancing Superpixel Segmentation and Classification Using Deep Learning Approaches

Accurate building detection is a critical task in urban development and digital city mapping. However, current models for high-resolution remote sensing images are still facing challenges due to complex object characteristics similarities appearance. To address this issue, paper proposes novel algorithm based on in-depth feature extraction classification of adaptive superpixel shredding. The proposed approach consists four main steps: image segmentation into homogeneous superpixels using modified Simple Linear Iterative Clustering (SLIC), an variational auto-encoder (VAE) scale the training testing data collection, identification classes (buildings, roads, trees, shadows) extracted as input Convolutional Neural Network (CNN), shapes through regional growth morphological operations. offers more stability identifying buildings with unclear boundaries, eliminating requirement extensive prior segmentation. It has been tested two datasets aerial from New Zealand region, demonstrating superior accuracy compared previous works average F1 score 98.83%. shows potential fast accurate monitoring planning, particularly areas.