Skip to main content

Table 5 The performance of the proposed network with different time step size values on KITTI 2015 test dataset

From: Refinement of matching costs for stereo disparities using recurrent neural networks

Time steps

Input sizes

Param. counts

B (%) > 5px

B (%) > 4px

B (%) > 3px

B (%) > 2px

1

192

164,288

9.24

9.95

11.24

14.17

2

96

188,864

9.10

9.88

11.23

13.83

3

64

229,824

9.31

10.04

11.24

13.85

4

48

274,880

9.16

9.90

11.15

13.93

8

24

465,344

9.04

9.80

11.13

13.92

12

16

659,904

9.02

9.76

10.95

13.37