Akademik Veri Yönetim Sistemi
28th Signal Processing and Communications Applications Conference, SIU 2020, Gaziantep, Türkiye, 5 - 07 Ekim 2020
© 2020 IEEE.Road defects affects driving as manner of both comfort and safety. For this reason, road damage detection studies are carried out continuously and roads are repaired periodically. The need for different equipment for road damage detection works and the installation, repair, maintenance of these equipments constitute extra cost. That's why, researchers are trying to detect road damages with the camera. In the studies conducted in the literature, road images are usually taken from the top, images are quality and cracks are clear.In this study, a new system designed for road damage detection that is designed and implemented as a simple and easy to use solution. The proposed system can classify when there are factors such as light, glass reection, and detect road damage without considering other objects entering the frame. Our convolutional neural network model, which was trained in accordance with these conditions, achieved 96% accuracy in damage detection from in-vehicle images.