Table 4.
Performance of different models in locating rice lodging areas
| Model | Parameters (M) | FLOPs (G) | PA (%) | MPA (%) | mIoU (%) |
|---|---|---|---|---|---|
| DeepLabV3 + MobileNetV2 | 5.81 | 2.64 | 93.3 | 93.3 | 88.2 |
| U2Net | 44.04 | 15.13 | 95.6 | 95.2 | 91.7 |
| PSPNet-ResNet50 | 46.7 | 5.92 | 95.2 | 95.1 | 91.2 |
| HRNet | 65.85 | 9.39 | 95.9 | 95.9 | 92.4 |
| ConvUNext | 3.50 | 2.87 | 95.6 | 95.5 | 91.9 |
| FA + ensemble + DeepLabV3+ | 67.35 | 4.78 | 94.7 | 94.6 | 90.4 |
| UConvNeXt | 5.83 | 4.04 | 95.2 | 95.2 | 91.3 |
| AUConvNeXt | 5.23 | 3.87 | 95.9 | 96.0 | 92.5 |
| AAUConvNeXt | 4.87 | 3.69 | 96.3 | 96.3 | 93.2 |