Detecting road anomalies early is crucial to prevent accidents and vehicle damage. Our model uses artificial intelligence techniques to automatically detect different types of anomalies, such as potholes, manholes, cracks, speed bumps, and others, in a very short time. It utilizes drones to capture videos and images. Various artificial intelligence techniques, including graph segmentation, graph similarity, and dynamic programming techniques are used for road segmentation, region proposals, anomaly detection, and classification of the detected anomalies. By combining these algorithms considerable results are achieved in terms of speed and accuracy; such that it generates around 90% regions less than the selective search model. Also, the MAP is increased considerably using the proposed model.