Table 5 The role of SRM and SIM of our method on the results.

From: Spatially adaptive interaction network for semantic segmentation of high-resolution remote sensing images

Dataset

Model

Spatial refine module

Spatial interation module

mIoU

DeepGlobe

Baseline

\(\times\)

\(\times\)

71.6

Ours

\(\checkmark\)

\(\times\)

65.7

\(\times\)

\(\checkmark\)

72.4

\(\checkmark\)

\(\checkmark\)

77.8

Vaihingen

Baseline

\(\times\)

\(\times\)

78.7

Ours

\(\checkmark\)

\(\times\)

75.6

\(\times\)

\(\checkmark\)

78.9

\(\checkmark\)

\(\checkmark\)

82.5

Postdam

Baseline

\(\times\)

\(\times\)

79.3

Ours

\(\checkmark\)

\(\times\)

70.3

\(\times\)

\(\checkmark\)

79.6

\(\checkmark\)

\(\checkmark\)

88.3

  1. Significant values are in bold.