Semantic Segmentation on LandCover dataset
88.02mIoUEnsemble UNet
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| Ensemble UNetType=Hybrid, Param. (M)=193.539, GFLOPs=141.26, log(Param.)=2.287, Input image size=512 x 512, Model Components=UNet MaxVIT-S, UNet ConvFormer-M36, and UNet EfficientNet-B72026.02 | 88.02 | 7.52 | 2.33 | 5.04 | |
| DAS-SKType=CNN, Param. (M)=10.678, GFLOPs=11.25, log(Param.)=1.028, Input image size=512 x 5122026.02 | 86.25 | 5.75 | 49.7 | 40.05 | |
| MA-DBFANType=Hybrid, Param. (M)=34.595, GFLOPs=853.3, log(Param.)=1.539, Input image size=512 x 5122026.02 | 85.3 | 4.8 | 0.37 | 0.23 | |
| SegFormer MiT-B2Type=Transformer, Param. (M)=27.461, GFLOPs=43.18, log(Param.)=1.439, Input image size=512 x 5122026.02 | 84.4 | 3.9 | 6.28 | 23.54 | |
| Diff-HRNetType=CNN, Param. (M)=68.867, GFLOPs=86.25, log(Param.)=1.838, Input image size=512 x 5122026.02 | 84.22 | 3.72 | 2.35 | 16.33 | |
| UNetType=CNN, Param. (M)=24.436, GFLOPs=31.41, log(Param.)=1.388, Input image size=512 x 5122026.02 | 83.4 | 2.9 | 6.65 | 54.34 | |
| DeepLabV3 ResNet101Type=CNN, Param. (M)=58.626, GFLOPs=241.14, log(Param.)=1.768, Input image size=512 x 5122026.02 | 83 | 2.5 | 0.59 | 9.59 | |
| TransUNetType=Hybrid, Param. (M)=91.404, GFLOPs=165.14, log(Param.)=1.961, Input image size=512 x 5122026.02 | 82.9 | 2.4 | 0.74 | 8.16 | |
| HRNetType=CNN, Param. (M)=11.458, GFLOPs=20.39, log(Param.)=1.059, Input image size=512 x 5122026.02 | 82.8 | 2.3 | 10.65 | 25.46 | |
| BiSeNetType=CNN, Param. (M)=11.878, GFLOPs=12.46, log(Param.)=1.075, Input image size=512 x 5122026.02 | 80.5 | — | — | 126.65 |