In Japan, we are trying to maintenance civil infrastructures to keep a next society sustainability. There is a lack of human resources involved in maintenance expert and Ministry of Land, Infrastructure, Transport and Tourism said to change the how to work called by i-Construction. It means that all process to concerned with civil infrastructure is improved efficiency. However, a local government has a lot of road routs rather than national government. In the survey of road pavement by national agencies, the road pavement is surveyed by a sophisticated system such as MCI ,MMS and so on. But local government don't have enough budget and experts. Furthermore, road specification is a local network. Turning to the information and mechanical engineering field, AI technology and motion senso are quickly developing. In this research, we are aiming to be efficient to maintain a road pavement condition for the local government administrator. We are developing to damaged detection due to roughness index of a road pavement using Deep Learning from video data in the car. And then improving damaged detection of crack and combining with exiting system which is a road pavement evaluation system which is monitoring to a motion sensor and video camera.