Machine Learning Based Pavement Crack Identification

AIM

  • To develop an automated pavement crack detection and classification algorithm
  • To create a faster and cheaper pavement management system

CURRENT STATE OF THE RESEARCH

  • A novel system based on image processing analyze of images taken from unmanned aerial vehicles and a machine learning algorithm, Support Vector Machines,  is developed.
  • The proposed system can be used for determining and classifying: (i) longitudinal, (ii) transverse, and (iii) alligator cracks. In addition, it can detect non-cracks on the pavement surface
  • Geographic Information System integration is performed by positioning classified crack images into Google Maps.