Table 6 Performance of the proposed method on the NYU dataset using different encoders.

From: LapUNet: a novel approach to monocular depth estimation using dynamic laplacian residual U-shape networks

Encoder

Model Size

Abs Rel

Log 10

RMSE

δ < 1.25

δ < 1.252

δ < 1.253

Low is better

High is better

MobileNetV250

19.3 M

0.121

0.562

0.565

0.856

0.943

0.975

VGG1951

217.8 M

0.113

0.513

0.521

0.878

0.951

0.979

ResNet-10116

60.2 M

0.096

0.042

0.426

0.811

0.981

0.989

DenseNet-16152

44.7 M

0.094

0.040

0.420

0.915

0.982

0.991

ResNeXt-10117

73.3 M

0.089

0.035

0.406

0.923

0.987

0.995

Back to article page