doi: 10.22712/SUSB.20210008

Accelerating multi-class defect detection of building façades using knowledge distillation of DCNN-based model