Automatic road anomaly detection using road image and motion sensor data

Crack detection, which is very important in different structures such as bridges, dams, reinforced concrete structures and highways, can be detected by image processing and classification. Detection of cracks on highways reduces traffic density, prevents accidents, eases the workload of road inspectors and provides better road maintenance. The possibility of human error is high due to a number of reasons such as manual inspection of cracks and defects found in asphalt, fatigue, irresponsible inspection, poor eye vision. Image processing and Machine Learning techniques have been used in recent years for the detection and analysis of cracks. These techniques are focused on identification and classification and automatic realization. Continuous mapping of road real conditions allows for the maintenance and management of adequate infrastructure along with a consistent resource allocation. Therefore, informing road users about the quality of infrastructure and getting information from the same users has become the new frontier of mobile device apps for navigation (like a private social network for safety road) for safe driving. In this study, a useful automatic detection system has been implemented for monitoring road surface quality and full road scene inventory images. The main goal is to find the optimal way to detect cracks found in roads and concrete structures. The applications focus on crack detection from images using structures such as VGG16 architecture, Convolutional Neural Network and Gabor Filter.