Skip to main content

Table 4 The recognition accuracies of the 23 categories for these five methods to combine these knowledge-based classifiers

From: Ensemble feature learning for material recognition with convolutional neural networks

 

Category

 
 

Brick

Carpet

Ceramic

Fabric

Foliage

Food

Glass

 

Vote

0.852

0.888

0.784

0.672

0.904

0.916

0.776

 

Stacking

0.828

0.896

0.700

0.588

0.876

0.852

0.712

 

Mean

0.852

0.884

0.768

0.684

0.908

0.916

0.792

 

Max

0.848

0.856

0.772

0.668

0.908

0.920

0.804

 

Weight

0.840

0.876

0.764

0.664

0.908

0.920

0.808

 
 

Category

 

Hair

Leather

Metal

Mirror

Other

Painted

Paper

Plastic

Vote

0.932

0.844

0.672

0.740

0.820

0.848

0.800

0.620

Stacking

0.9

0.832

0.652

0.748

0.768

0.836

0.716

0.524

Mean

0.94

0.840

0.680

0.744

0.832

0.840

0.816

0.640

Max

0.932

0.836

0.680

0.740

0.828

0.840

0.832

0.652

Weight

0.94

0.844

0.676

0.736

0.828

0.844

0.840

0.668

 

Category

 

Polished stone

Skin

Sky

Stone

Tile

Wallpaper

Water

Wood

Vote

0.812

0.904

0.980

0.796

0.712

0.860

0.924

0.672

Stacking

0.764

0.884

0.976

0.752

0.688

0.852

0.904

0.588

Mean

0.832

0.900

0.980

0.820

0.716

0.852

0.940

0.684

Max

0.836

0.900

0.980

0.828

0.720

0.844

0.940

0.692

Weight

0.824

0.912

0.980

0.840

0.720

0.852

0.932

0.684