combine dask ops

docker/requirements.txt

@@ -11,7 +11,7 @@ odc-dscache>=0.2.3
 odc-stac @ git+https://github.com/opendatacube/odc-stac@69bdf64
 
 # odc-stac is in PyPI
-odc-stats[ows] @ git+https://github.com/opendatacube/odc-stats@d7307d9
+odc-stats[ows] @ git+https://github.com/opendatacube/odc-stats@a9bdd82
 
 # For ML
 tflite-runtime

odc/stats/plugins/lc_treelite_cultivated.py

@@ -227,6 +227,12 @@ def generate_features(input_block, bands_indices):
     return output_block
 
 
+def cultivated_predict(input_block, bands_indices):
+    feature_block = generate_features(input_block, bands_indices)
+    cc = mask_and_predict(feature_block, ptype="categorical", nodata=NODATA)
+    return cc
+
+
 class StatsCultivatedClass(StatsMLTree):
     NAME = "ga_ls_cultivated"
     SHORT_NAME = NAME
@@ -240,23 +246,10 @@ def measurements(self) -> Tuple[str, ...]:
 
     def predict(self, input_array):
         bands_indices = dict(zip(self.input_bands, np.arange(len(self.input_bands))))
-        input_features = da.map_blocks(
-            generate_features,
+        cc = da.map_blocks(
+            cultivated_predict,
             input_array,
             bands_indices,
-            chunks=(
-                input_array.chunks[0],
-                input_array.chunks[1],
-                15 + len(bands_indices) - bands_indices["bcdev"] - 1,
-            ),
-            dtype="float32",
-            name="generate_features",
-        )
-        cc = da.map_blocks(
-            mask_and_predict,
-            input_features,
-            ptype="categorical",
-            nodata=NODATA,
             drop_axis=-1,
             dtype="float32",
             name="cultivated_predict",