Instructions to use ibm-nasa-geospatial/Prithvi-EO-2.0-300M-TL-Sen1Floods11 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- TerraTorch
How to use ibm-nasa-geospatial/Prithvi-EO-2.0-300M-TL-Sen1Floods11 with TerraTorch:
from terratorch.registry import BACKBONE_REGISTRY model = BACKBONE_REGISTRY.build("ibm-nasa-geospatial/Prithvi-EO-2.0-300M-TL-Sen1Floods11") - Notebooks
- Google Colab
- Kaggle
Commit ·
e54b97f
1
Parent(s): 70c9404
Fix rename weight keys and fix inference
Browse files- Prithvi-EO-V2-300M-TL-Sen1Floods11.pt +2 -2
- config.yaml +0 -21
- inference.py +2 -2
- requirements.txt +1 -1
Prithvi-EO-V2-300M-TL-Sen1Floods11.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3675e9c2b52547de8ff8a19f4881c28573e6d4d2f0805d866f4fc48c1e517d60
|
| 3 |
+
size 1276843350
|
config.yaml
CHANGED
|
@@ -101,13 +101,6 @@ data:
|
|
| 101 |
- SWIR_1
|
| 102 |
- SWIR_2
|
| 103 |
train_transform:
|
| 104 |
-
- class_path: albumentations.Resize
|
| 105 |
-
init_args:
|
| 106 |
-
height: 448
|
| 107 |
-
width: 448
|
| 108 |
-
interpolation: 1
|
| 109 |
-
always_apply: false
|
| 110 |
-
p: 1
|
| 111 |
- class_path: albumentations.RandomCrop
|
| 112 |
init_args:
|
| 113 |
height: 224
|
|
@@ -128,26 +121,12 @@ data:
|
|
| 128 |
always_apply: true
|
| 129 |
p: 1.0
|
| 130 |
val_transform:
|
| 131 |
-
- class_path: albumentations.Resize
|
| 132 |
-
init_args:
|
| 133 |
-
height: 448
|
| 134 |
-
width: 448
|
| 135 |
-
interpolation: 1
|
| 136 |
-
always_apply: false
|
| 137 |
-
p: 1
|
| 138 |
- class_path: albumentations.pytorch.ToTensorV2
|
| 139 |
init_args:
|
| 140 |
transpose_mask: false
|
| 141 |
always_apply: true
|
| 142 |
p: 1.0
|
| 143 |
test_transform:
|
| 144 |
-
- class_path: albumentations.Resize
|
| 145 |
-
init_args:
|
| 146 |
-
height: 448
|
| 147 |
-
width: 448
|
| 148 |
-
interpolation: 1
|
| 149 |
-
always_apply: false
|
| 150 |
-
p: 1
|
| 151 |
- class_path: albumentations.pytorch.ToTensorV2
|
| 152 |
init_args:
|
| 153 |
transpose_mask: false
|
|
|
|
| 101 |
- SWIR_1
|
| 102 |
- SWIR_2
|
| 103 |
train_transform:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 104 |
- class_path: albumentations.RandomCrop
|
| 105 |
init_args:
|
| 106 |
height: 224
|
|
|
|
| 121 |
always_apply: true
|
| 122 |
p: 1.0
|
| 123 |
val_transform:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 124 |
- class_path: albumentations.pytorch.ToTensorV2
|
| 125 |
init_args:
|
| 126 |
transpose_mask: false
|
| 127 |
always_apply: true
|
| 128 |
p: 1.0
|
| 129 |
test_transform:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 130 |
- class_path: albumentations.pytorch.ToTensorV2
|
| 131 |
init_args:
|
| 132 |
transpose_mask: false
|
inference.py
CHANGED
|
@@ -185,7 +185,7 @@ def run_model(input_data, temporal_coords, location_coords, model, datamodule, i
|
|
| 185 |
for x in windows:
|
| 186 |
# Apply standardization
|
| 187 |
x = datamodule.test_transform(image=x.squeeze().numpy().transpose(1,2,0))
|
| 188 |
-
x = datamodule.aug(x['image']
|
| 189 |
|
| 190 |
with torch.no_grad():
|
| 191 |
x = x.to(model.device)
|
|
@@ -237,7 +237,7 @@ def main(
|
|
| 237 |
# Load model ---------------------------------------------------------------------------------
|
| 238 |
|
| 239 |
lightning_model = LightningInferenceModel.from_config(config, checkpoint)
|
| 240 |
-
img_size = 512 # Size of Sen1Floods11
|
| 241 |
|
| 242 |
# Loading data ---------------------------------------------------------------------------------
|
| 243 |
|
|
|
|
| 185 |
for x in windows:
|
| 186 |
# Apply standardization
|
| 187 |
x = datamodule.test_transform(image=x.squeeze().numpy().transpose(1,2,0))
|
| 188 |
+
x = datamodule.aug(x)['image']
|
| 189 |
|
| 190 |
with torch.no_grad():
|
| 191 |
x = x.to(model.device)
|
|
|
|
| 237 |
# Load model ---------------------------------------------------------------------------------
|
| 238 |
|
| 239 |
lightning_model = LightningInferenceModel.from_config(config, checkpoint)
|
| 240 |
+
img_size = 512 # Size of Sen1Floods11 training
|
| 241 |
|
| 242 |
# Loading data ---------------------------------------------------------------------------------
|
| 243 |
|
requirements.txt
CHANGED
|
@@ -3,4 +3,4 @@ torchvision
|
|
| 3 |
timm
|
| 4 |
einops
|
| 5 |
rasterio
|
| 6 |
-
|
|
|
|
| 3 |
timm
|
| 4 |
einops
|
| 5 |
rasterio
|
| 6 |
+
terratorch==0.99.8
|