improved_sen1flood11_lr

README.md

@@ -15,7 +15,7 @@ metrics:
 - IoU
 ---
 ### Model and Inputs
-The pretrained [Prithvi-EO-1.0-100m](https://huggingface.co/ibm-nasa-geospatial/Prithvi-100M/blob/main/README.md) model is finetuned to segment the extent of floods on Sentinel-2 images from the [Sen1Floods11 dataset](https://github.com/cloudtostreet/Sen1Floods11). 
+The pretrained [Prithvi-100m](https://huggingface.co/ibm-nasa-geospatial/Prithvi-100M/blob/main/README.md) model is finetuned to segment the extent of floods on Sentinel-2 images from the [Sen1Floods11 dataset](https://github.com/cloudtostreet/Sen1Floods11). 
 
 The dataset consists of 446 labeled 512x512 chips that span all 14 biomes, 357 ecoregions, and 6 continents of the world across 11 flood events. The benchmark associated to Sen1Floods11 provides results for fully convolutional neural networks trained in various input/labeled data setups, considering Sentinel-1 and Sentinel-2 imagery. 
 
@@ -28,7 +28,7 @@ We extract the following bands for flood mapping:
 5. SWIR 1
 6. SWIR 2
 
-Labels represent no water (class 0), water/flood (class 1), and no data/clouds (class -1).
+Labels represent no water (class 0), water/flood (class 1), and no data/clouds (class 2).
 
 The Prithvi-100m model was initially pretrained using a sequence length of 3 timesteps. Based on the characteristics of this benchmark dataset, we focus on single-timestamp segmentation. This demonstrates that our model can be utilized with an arbitrary number of timestamps during finetuning.
 

sen1floods11_Prithvi_100M.py

@@ -74,7 +74,8 @@ train_pipeline = [
         type="LoadGeospatialImageFromFile",
         to_float32=False,
         nodata=image_nodata,
-        nodata_replace=image_nodata_replace
+        nodata_replace=image_nodata_replace,
+        channels_last=False
     ),
     dict(
         type="LoadGeospatialAnnotations",
@@ -106,7 +107,8 @@ test_pipeline = [
         type="LoadGeospatialImageFromFile",
         to_float32=False,
         nodata=image_nodata,
-        nodata_replace=image_nodata_replace
+        nodata_replace=image_nodata_replace,
+        channels_last=False
     ),
     dict(type="BandsExtract", bands=bands),
     dict(type="ConstantMultiply", constant=constant),
@@ -224,9 +226,9 @@ ce_weights = [0.3, 0.7]
 model = dict(
     type="TemporalEncoderDecoder",
     frozen_backbone=False,
-    pretrained=pretrained_weights_path,
     backbone=dict(
         type="TemporalViTEncoder",
+        pretrained=pretrained_weights_path,
         img_size=img_size,
         patch_size=patch_size,
         num_frames=num_frames,