A hybrid deep transfer learning approach for road extraction from high-resolution satellite images based on D-LinkNet and fuzzy systems

Key Points

• The hybrid model for road extraction yielded superior accuracy using deep transfer learning techniques, improving performance.
• Key metrics showed that the combination of D-LinkNet and fuzzy systems resulted in enhanced image processing capabilities.
• Analysis employed advanced algorithms for extracting detailed road features from satellite imagery, addressing traditional limitations.
• Further validation on diverse datasets is needed to confirm the robustness of this innovative approach.