Researchers have developed ResAF-Net, a novel deep learning framework for detecting trees and mapping agricultural areas using satellite imagery, specifically designed for resource-constrained regions like Palestine. The model integrates a ResNet-50 encoder with an anchor-free detection head and a self-attention module to enhance localization accuracy in complex landscapes. Tested on the MillionTrees benchmark, ResAF-Net demonstrated high recall and competitive mean average precision, showcasing its effectiveness for agricultural inventorying. The framework has been deployed in a GIS application to support data-driven analysis at various geographical scales. AI
Summary written by gemini-2.5-flash-lite from 1 source. How we write summaries →
IMPACT Provides a scalable AI solution for agricultural monitoring in data-scarce regions, enabling data-driven land-use planning.
RANK_REASON This is a research paper detailing a new model and its application.