fix(earthview.py): compatibility with latest EarthView, with updated metadata

earthview.py

@@ -12,7 +12,7 @@ DATASET = "satellogic/EarthView"
 
 sets = {
     "satellogic": {
-        "shards" : 3676,
+        "shards" : 7863,
     },
     "sentinel_1": {
         "shards" : 1763,
@@ -43,7 +43,11 @@ def load_dataset(subset, dataset="satellogic/EarthView", split="train", shards =
     if shards is None:
         data_files = None
     else:
-        data_files = [f"{path}/{split}-{shard:05d}-of-{nshards:05d}.parquet" for shard in shards]
+        if subset == "satellogic":
+            shards = {1000*shard//nshards for shard in shards}
+            data_files = [f"{path}/satellogic_v2-{shard}/{split}-*.parquet" for shard in shards]
+        else:
+            data_files = [f"{path}/{split}-{shard:05d}-of-{nshards:05d}.parquet" for shard in shards]
         data_files = {split: data_files}
 
     ds = _load_dataset(

utils.py

@@ -15,8 +15,7 @@ def lat_lon_mid_pixel(item, subset: str):
     metadata = get_metadata(item)
     if subset == "satellogic":
         crs = metadata["crs"][0]
-        # each image has bounds but they should coincide
-        bounds_crs = metadata["bounds"][0]
+        bounds_crs = metadata["bounds"]
     elif subset == "sentinel_1":
         crs = metadata["crs"]
         bounds_crs = metadata["coordinates"][0]
@@ -28,6 +27,10 @@ def lat_lon_mid_pixel(item, subset: str):
             bounds_crs[2][0],
             bounds_crs[2][1],
         )
+    # elif subset == "neon":
+    #     crs = metadata["epsg"]
+    #     bounds_crs = metadata["bounds"]
+
     else:
         raise ValueError("subset not known")