Table 3 Loss comparison of different models.

From: Dense monocular depth estimation for stereoscopic vision based on pyramid transformer and multi-scale feature fusion

Method

Epoch

Training loss

Validation loss

ViT-32

10

1.79

2.85

20

1.45

2.59

30

1.29

2.41

45

1.20

2.29

ViT-16

10

1.55

2.62

20

1.31

2.36

30

1.20

2.25

45

1.14

2.20

Hybrid

10

1.27

2.34

20

1.17

2.21

30

1.14

2.16

45

1.07

2.12

PVT

10

1.21

2.31

20

1.12

2.16

30

1.10

2.12

45

1.06

2.08

SPT-Depth

10

1.14

2.16

20

1.08

2.07

30

1.06

2.04

45

1.04

2.00

  1. Significant values are in [bold].
Back to article page