Road Damage Detection System Using Canny Edge Detection Algorithm for Time Efficiency in Road Condition Survey

Abstract

Land transportation is still the main focus of transportation in Indonesia. With a total length of roads reaching hundreds of thousands of kilometers, it is necessary to monitor the health of the roads to ensure the roads can be traversed properly and immediately follow up if there are roads that are not suitable for passage. Currently, Dinas Bina Marga dan Penataan Ruang Provinsi Jawa Barat (unit that carries out government affairs in the field of public works and spatial planning including road sub-affairs, construction services sub-affairs, and spatial planning sub-affairsis) working with the Bandung Institute of Technology to develop the Survei Kondisi Perkerasan Jalan application to support the efficiency of Road Condition Survey activities. However, this application is still running semi-automatically with human intervention, one of which is the detection process. For that we need a solution in the form of detecting road damage automatically. This research aims to detect road damage automatically using the Canny Edge Detection algorithm. In operation, the system is capable of detecting road damage and selecting the damaged road area. The benefit of this research is to simplify the road damage classification process and time efficiency of road condition survey activities. The results of the tests carried out on video recordings of roads with a camera angle of 0 degrees using the Canny Edge Detection algorithm are 54.5% accuracy, 24.2% precision, 78.9% recall, and 37% F-score.