Road Damage Classification using Back Propagation Algorithm

Abstract

Industrial development in Indonesia impact to the road damage at several districts. Because the industrial transport which owned by some industry to distribute the product were heavy transport, so the road damage could not be avoided. So that, the government was expected to be able to detect the road damage earlier before many traffic accident in various districts happened. To support the government performance handling the road damage, the government has to be able to have classifications the road damage so the government does not wrong in handle the damage. As the technology development, the government could apply information technology to have the road damage classifications using computer. Backpropagation method was one of the neural networks method that able to do road damage classifications. With the digital image processing support, system would have good accuracy. Some of digital image processing method used for cropping, edge detection, and thresholding. The data used in this research was a photograph from the highway, where the data consisting of road damage image and a photograph of the road that there was no the damage. Then the data will have the image processing first prior to classifications by system that used backpropagation method. So that, in this research produced a system that could do road damage classifications with the accuracy about 84 %.