Table 3 Detection results of the trained model on sectional point cloud images

From: Intelligent Hbim modeling of traditional timber architectural heritage, integrating machine vision and architectural expertise

Class

Precision

Recall

mAP50

all

0.897

0.850

0.901

outer-YZ

0.974

0.969

0.989

outer-GZ

0.944

0.757

0.875

inner-ZZ

0.977

0.991

0.985

inner-JZ

1

0.997

0.995

other-GZ

0.936

0.749

0.863

scale

0.994

1

0.995

left-TF

0.954

0.823

0.911

right-TF

0.969

0.825

0.914

FZ

0.699

0.736

0.799

LGZ

0.981

0.958

0.992

ground

0.925

0.898

0.939

CF

0.842

0.837

0.889

GeZ

0.839

0.724

0.843

plate

0.913

0.885

0.929

roof-GZ

0.749

0.77

0.798

roof-CF

0.656

0.674

0.71

Back to article page