Update README.md
Browse files
README.md
CHANGED
|
@@ -32,4 +32,50 @@ This modelcard aims to be a base template for new models. It has been generated
|
|
| 32 |
- **Paper [optional]:** [More Information Needed]
|
| 33 |
- **Demo [optional]:** [More Information Needed]
|
| 34 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 35 |
## Uses
|
|
|
|
| 32 |
- **Paper [optional]:** [More Information Needed]
|
| 33 |
- **Demo [optional]:** [More Information Needed]
|
| 34 |
|
| 35 |
+
---
|
| 36 |
+
Table 1: Linear probing results on six classification tasks. All models are trained
|
| 37 |
+
for 50 epochs. The reported numbers are top-1 overall accuracy (OA). Missing values
|
| 38 |
+
are due to the inability of the model to adapt to this domain.
|
| 39 |
+
|
| 40 |
+
| Method | Backbone | m-bigearthnet | m-forestnet | m-brick-kiln | m-pv4ger | m-so2sat | m-eurosat |
|
| 41 |
+
|--------------------|-------------|---------------|-------------|--------------|----------|----------|-----------|
|
| 42 |
+
| **Fully Trained** | ViT-S | 66.0 | 53.8 | 98.1 | 97.6 | 57.5 | 97.3 |
|
| 43 |
+
| **Fully Trained** | SwinV2-T | 70.0 | 58.0 | 98.7 | 98.0 | 56.1 | 97.4 |
|
| 44 |
+
| **Fully Trained** | ConvNext-B | 69.1 | 56.8 | 98.9 | 98.0 | 58.1 | 97.7 |
|
| 45 |
+
| **rand. init.** | ViT-B | 52.9 | 41.5 | 84.5 | 91.3 | 38.3 | 85.7 |
|
| 46 |
+
| **MAE_Single [44]**| ViT-B | 63.6 | - | 88.9 | 92.2 | 50.0 | 88.9 |
|
| 47 |
+
| **OFA-Net [43]** | ViT-B | 65.0 | - | 94.7 | 93.2 | 49.4 | 91.9 |
|
| 48 |
+
| **SatMAE [25]** | ViT-B | 62.1 | - | 93.9 | - | 46.9 | 86.4 |
|
| 49 |
+
| **Scale-MAE [22]** | ViT-L | - | - | - | 96.9 | - | - |
|
| 50 |
+
| **GFM [21]** | Swin-B | - | - | - | 96.8 | - | - |
|
| 51 |
+
| **Cross-Scale MAE [23]** | ViT-B | - | - | - | 93.1 | - | - |
|
| 52 |
+
| **FG-MAE [24]** | ViT-B | 63.0 | - | 94.7 | - | 51.4 | 87.0 |
|
| 53 |
+
| **CROMA [27]** | ViT-B | 67.4 | - | 91.0 | - | 49.2 | 90.1 |
|
| 54 |
+
| **DOFA** | ViT-B | 65.7 | 50.9 | 95.8 | 96.9 | 55.1 | 93.9 |
|
| 55 |
+
| **DOFA** | ViT-L | 67.5 | 54.6 | 96.9 | 97.3 | 60.1 | 97.1 |
|
| 56 |
+
|
| 57 |
+
|
| 58 |
+
|
| 59 |
+
Partial fine-tuning results on six segmentation tasks. All models are
|
| 60 |
+
trained with a frozen backbone for 20 epochs. Reported numbers are mean intersection
|
| 61 |
+
over union (mIoU). Missing values are due to the inability of the model to adapt to
|
| 62 |
+
this domain.
|
| 63 |
+
|
| 64 |
+
| Method | Backbone | m-pv4ger-seg | m-nz-cattle | m-NeonTree | m-cashew-plant | m-SA-crop | m-chesapeake |
|
| 65 |
+
|--------------------|-------------|--------------|-------------|------------|----------------|-----------|--------------|
|
| 66 |
+
| **DeepLabv3** | ResNet101 | 93.4 | 67.6 | 53.9 | 48.6 | 30.4 | 62.1 |
|
| 67 |
+
| **U-Net** | ResNet101 | 94.1 | 80.5 | 56.6 | 46.6 | 29.9 | 70.8 |
|
| 68 |
+
| **rand. init.** | ViT-B | 81.7 | 74.1 | 51.7 | 32.4 | 29.0 | 47.1 |
|
| 69 |
+
| **MAE_Single [44]**| ViT-B | 88.4 | 76.4 | 53.0 | 40.7 | 30.7 | 51.9 |
|
| 70 |
+
| **OFA-Net [43]** | ViT-B | 89.4 | 77.6 | 53.3 | 47.9 | 31.9 | 54.5 |
|
| 71 |
+
| **Scale-MAE [22]** | ViT-L | 83.5 | 76.5 | 51.0 | - | - | 61.0 |
|
| 72 |
+
| **GFM [21]** | Swin-B | 92.0 | 75.0 | 51.1 | - | - | 63.8 |
|
| 73 |
+
| **Cross-Scale MAE [23]** | ViT-B | 83.2 | 77.9 | 52.1 | - | - | 52.3 |
|
| 74 |
+
| **CROMA [27]** | ViT-B | - | - | - | 30.1 | 31.4 | - |
|
| 75 |
+
| **FG-MAE [24]** | ViT-B | - | - | - | 40.8 | 30.6 | - |
|
| 76 |
+
| **DOFA** | ViT-B | 94.5 | 81.4 | 58.8 | 51.5 | **33.0** | 65.3 |
|
| 77 |
+
| **DOFA** | ViT-L | 95.0 | 81.8 | 59.4 | **56.9** | **32.1** | 66.3 |
|
| 78 |
+
|
| 79 |
+
---
|
| 80 |
+
|
| 81 |
## Uses
|