Each image of pavement of these datasets has a ground-truth associated. Moreover these datasets are separated in three categories :