48Mouad/road-damage-detection-morocco

Multi-class road damage detection using deep learning, with model benchmarking, damaged-area estimation, and domain generalization on RDD2022 and Moroccan road data. # Road Damage Detection Morocco Multi-class road damage detection, model benchmarking, damaged-area estimation, and domain generalization using RDD2022 and Moroccan road data. ## Objectives - Audit and harmonize multiple road damage datasets - Train and compare several object detection models - Evaluate model generalization on Moroccan roads - Generate scientific figures and performance indicators - Estimate apparent damaged-road coverage - Identify the most suitable model for operational deployment ## Target Road Damage Categories - Longitudinal cracks - Transverse cracks - Alligator cracks - Potholes - Rutting - Depressions - Raveling - Bleeding - Degraded road markings ## Repository Structure ```text configs/ Project configurations data/ Local datasets, excluded from Git notebooks/ Experimental notebooks src/ Source code scripts/ Execution scripts tests/ Automated tests outputs/ Figures, tables, reports and predictions