Developed an advanced AI model to detect road anomalies (potholes, cracks, surface damage) in real time, enhancing road safety and infrastructure maintenance. Built using a TripletNet architecture with ResNet18 backbone, the model flags hazards when embedding distance >2.5. Trained and validated on the RDD2022 dataset (47k+ images from 6 countries), achieving strong generalization across global conditions. Contributed to ZewailCity’s mission of leveraging AI for safer, smarter transportation systems.
