Merge pull request #188 from WorldCereal/simplify-majority-vote

get rid of conf_threshold parameter as it may introduce "undecided" p…

notebooks/worldcereal_v1_demo_custom_cropland.ipynb

@@ -490,7 +490,6 @@
     "postprocess_method = \"majority_vote\"\n",
     "# Additiona parameters for the majority vote method:\n",
     "kernel_size = 3 # default = 5\n",
-    "conf_threshold = 60 # default = 30\n",
     "# Do you want to save the intermediate results (before applying the postprocessing)\n",
     "save_intermediate = True #default is False\n",
     "# Do you want to save all class probabilities in the final product? (default is False)\n",
@@ -499,7 +498,6 @@
     "postprocess_parameters = PostprocessParameters(enable=postprocess_result,\n",
     "                                               method=postprocess_method,\n",
     "                                               kernel_size=kernel_size,\n",
-    "                                               conf_threshold=conf_threshold,\n",
     "                                               save_intermediate=save_intermediate,\n",
     "                                               keep_class_probs=keep_class_probs)\n",
     "\n",

notebooks/worldcereal_v1_demo_custom_croptype.ipynb

@@ -508,7 +508,6 @@
     "postprocess_method = \"majority_vote\"\n",
     "# Additiona parameters for the majority vote method:\n",
     "kernel_size = 5 # default = 5\n",
-    "conf_threshold = 30 # default = 30\n",
     "# Do you want to save the intermediate results (before applying the postprocessing)\n",
     "save_intermediate = True #default is False\n",
     "# Do you want to save all class probabilities in the final product? (default is False)\n",
@@ -517,7 +516,6 @@
     "postprocess_parameters = PostprocessParameters(enable=postprocess_result,\n",
     "                                               method=postprocess_method,\n",
     "                                               kernel_size=kernel_size,\n",
-    "                                               conf_threshold=conf_threshold,\n",
     "                                               save_intermediate=save_intermediate,\n",
     "                                               keep_class_probs=keep_class_probs)\n",
     "\n",

notebooks/worldcereal_v1_demo_default_cropland_EXTENDED.ipynb

@@ -222,7 +222,6 @@
     "postprocess_method = \"majority_vote\"\n",
     "# Additiona parameters for the majority vote method:\n",
     "kernel_size = 3 # default = 5\n",
-    "conf_threshold = 60 # default = 30\n",
     "# Do you want to save the intermediate results (before applying the postprocessing)\n",
     "save_intermediate = True #default is False\n",
     "# Do you want to save all class probabilities in the final product? (default is False)\n",
@@ -231,7 +230,6 @@
     "postprocess_parameters = PostprocessParameters(enable=postprocess_result,\n",
     "                                               method=postprocess_method,\n",
     "                                               kernel_size=kernel_size,\n",
-    "                                               conf_threshold=conf_threshold,\n",
     "                                               save_intermediate=save_intermediate,\n",
     "                                               keep_class_probs=keep_class_probs)\n",
     "\n",

src/worldcereal/openeo/postprocess.py

@@ -33,11 +33,10 @@ def majority_vote(
         base_labels: xr.DataArray,
         max_probabilities: xr.DataArray,
         kernel_size: int,
-        conf_threshold: int,
     ) -> xr.DataArray:
         """Majority vote is performed using a sliding local kernel.
-        For each pixel, the voting of a final class is done from
-        neighbours values weighted with the confidence threshold.
+        For each pixel, the voting of a final class is done by counting
+        neighbours values.
         Pixels that have one of the specified excluded values are
         excluded in the voting process and are unchanged.
 
@@ -55,8 +54,6 @@ def majority_vote(
             The original probabilities of the winning class (ranging between 0 and 100).
         kernel_size : int
             The size of the kernel used for the neighbour around the pixel.
-        conf_threshold : int
-            Pixels under this confidence threshold do not count into the voting process.
 
         Returns
         -------
@@ -93,9 +90,6 @@ def majority_vote(
             # Take the binary mask of the interest class, and multiply by the probabilities
             class_mask = ((prediction == cls_value) * probability).astype(np.uint16)
 
-            # Sets to 0 the class scores where the threshold is lower
-            class_mask[probability <= conf_threshold] = 0
-
             # Set to 0 the class scores where the label is excluded
             for excluded_value in cls.EXCLUDED_VALUES:
                 class_mask[prediction == excluded_value] = 0
@@ -156,7 +150,7 @@ def majority_vote(
         # Setting excluded values back to their original values
         for excluded_value in cls.EXCLUDED_VALUES:
             aggregated_predictions[prediction == excluded_value] = excluded_value
-            aggregated_probabilities[prediction == excluded_value] = cls.NODATA
+            aggregated_probabilities[prediction == excluded_value] = excluded_value
 
         return xr.DataArray(
             np.stack((aggregated_predictions, aggregated_probabilities)),
@@ -286,13 +280,11 @@ def execute(self, inarr: xr.DataArray) -> xr.DataArray:
         elif self._parameters.get("method") == "majority_vote":
 
             kernel_size = self._parameters.get("kernel_size")
-            conf_threshold = self._parameters.get("conf_threshold")
 
             new_labels = PostProcessor.majority_vote(
                 inarr.sel(bands="classification"),
                 inarr.sel(bands="probability"),
                 kernel_size=kernel_size,
-                conf_threshold=conf_threshold,
             )
 
             # Append the per-class probabalities if required

src/worldcereal/parameters.py

@@ -168,8 +168,6 @@ class PostprocessParameters(BaseModel):
         The method to use for postprocessing. Must be one of ["smooth_probabilities", "majority_vote"]
     kernel_size: int (default=5)
         Used for majority vote postprocessing. Must be smaller than 25.
-    conf_threshold: int (default=30)
-        Used for majority vote postprocessing. Must be between 0 and 100.
     save_intermediate: bool (default=False)
         Whether to save intermediate results (before applying the postprocessing).
         The intermediate results will be saved in the GeoTiff format.
@@ -180,7 +178,6 @@ class PostprocessParameters(BaseModel):
     enable: bool = Field(default=True)
     method: str = Field(default="smooth_probabilities")
     kernel_size: int = Field(default=5)
-    conf_threshold: int = Field(default=30)
     save_intermediate: bool = Field(default=False)
     keep_class_probs: bool = Field(default=False)
 
@@ -213,9 +210,5 @@ def check_parameters(self):
                 raise ValueError(
                     f"Kernel size must be smaller than 25, got {self.kernel_size}"
                 )
-            if self.conf_threshold < 0 or self.conf_threshold > 100:
-                raise ValueError(
-                    f"Confidence threshold must be between 0 and 100, got {self.conf_threshold}"
-                )
 
         return self

tests/worldcerealtests/test_postprocessing.py

@@ -50,7 +50,6 @@ def test_cropland_postprocessing_majority_vote(WorldCerealCroplandClassification
             "lookup_table": lookup_table,
             "method": "majority_vote",
             "kernel_size": 7,
-            "conf_threshold": 30,
         },
     )
 
@@ -90,7 +89,6 @@ def test_croptype_postprocessing_majority_vote(WorldCerealCroptypeClassification
             "lookup_table": lookup_table,
             "method": "majority_vote",
             "kernel_size": 7,
-            "conf_threshold": 30,
         },
     )
 
@@ -103,7 +101,6 @@ def test_postprocessing_parameters():
         "enable": True,
         "method": "smooth_probabilities",
         "kernel_size": 5,
-        "conf_threshold": 30,
         "save_intermediate": False,
         "keep_class_probs": False,
     }
@@ -118,12 +115,6 @@ def test_postprocessing_parameters():
     with pytest.raises(ValueError):
         PostprocessParameters(**params)
 
-    # This one should fail with invalid conf_threshold
-    params["kernel_size"] = 5
-    params["conf_threshold"] = 101
-    with pytest.raises(ValueError):
-        PostprocessParameters(**params)
-
     # This one should fail with invalid method
     params["method"] = "test"
     with pytest.raises(ValueError):