Table 3 Per-class mIoU comparison on LoveDA dataset with other Models.
Method | Backbone | Background | Building | Road | Water | Barren | Forest | Agriculture | mIoU(%) |
|---|---|---|---|---|---|---|---|---|---|
DeepLabv3+7 | R50 | 0.43 | 0.51 | 0.52 | 0.74 | 0.11 | 0.44 | 0.59 | 47.60 |
Segmenter26 | ViT-T | 0.38 | 0.51 | 0.49 | 0.77 | 0.13 | 0.44 | 0.58 | 47.10 |
ABCNet27 | R50 | 0.53 | 0.62 | 0.52 | 0.62 | 0.30 | 0.42 | 0.47 | 49.80 |
BANet28 | Res-T | 0.54 | 0.62 | 0.51 | 0.65 | 0.27 | 0.44 | 0.48 | 50.15 |
UNetformer29 | R18 | 0.45 | 0.59 | 0.55 | 0.80 | 0.20 | 0.46 | 0.63 | 50.73 |
CM-UNet30 | R18 | 0.55 | 0.64 | 0.56 | 0.68 | 0.30 | 0.43 | 0.51 | 52.17 |
MSPNet | Starnet_s4 | 0.49 | 0.46 | 0.48 | 0.67 | 0.37 | 0.57 | 0.66 | 58.33 |