update the metrics calculation

Author: supathdhitalGEO

dist/fimeval-0.1.45-py3-none-any.whl

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "fimeval"
-version = "0.1.44"
+version = "0.1.45"
 description = "A Framework for Automatic Evaluation of Flood Inundation Mapping Predictions Evaluation"
 authors = [
     "Surface Dynamics Modeling Lab",

dist/fimeval-0.1.45.tar.gz

@@ -40,7 +40,6 @@ def evaluateFIM(
     Merged = []
     Unique = []
     FAR_values = []
-    Dice_values = []
 
     # Dynamically call the specified method
     method = globals().get(method)
@@ -255,7 +254,6 @@ def resize_image(
                 FPR,
                 merged,
                 FAR,
-                Dice,
             ) = evaluationmetrics(out_image1, out_image2_resized)
 
             # Append values to the lists
@@ -275,7 +273,6 @@ def resize_image(
             Merged.append(merged)
             Unique.append(unique_values)
             FAR_values.append(FAR)
-            Dice_values.append(Dice)
 
     results = {
         "CSI_values": csi_values,
@@ -294,7 +291,6 @@ def resize_image(
         # 'Merged': Merged,
         #  'Unique': Unique
         "FAR_values": FAR_values,
-        "Dice_values": Dice_values,
     }
     for candidate_idx, candidate_path in enumerate(candidate_paths):
         candidate_BASENAME = os.path.splitext(os.path.basename(candidate_path))[0]

poetry.lock

@@ -7,21 +7,21 @@ def evaluationmetrics(out_image1, out_image2):
     unique_values, counts = np.unique(merged, return_counts=True)
     class_pixel_counts = dict(zip(unique_values, counts))
     class_pixel_counts
-    TN = class_pixel_counts[1]
-    FP = class_pixel_counts[2]
-    FN = class_pixel_counts[3]
-    TP = class_pixel_counts[4]
-    TPR = TP / (TP + FN)
-    FNR = FN / (TP + FN)
-    Acc = (TP + TN) / (TP + TN + FP + FN)
-    Prec = TP / (TP + FP)
-    sen = TP / (TP + FN)
-    F1_score = 2 * (Prec * sen) / (Prec + sen)
-    CSI = TP / (TP + FN + FP)
-    POD = TP / (TP + FN)
-    FPR = FP / (FP + TN)
-    FAR = FP / (TP + FP)
-    Dice = 2 * TP / (2 * TP + FP + FN)
+    TN = class_pixel_counts.get(1,0)
+    FP = class_pixel_counts.get(2,0)
+    FN = class_pixel_counts.get(3,0)
+    TP = class_pixel_counts.get(4,0)
+    epsilon = 1e-8
+    TPR = TP / (TP + FN+epsilon)
+    FNR = FN / (TP + FN+epsilon)
+    Acc = (TP + TN) / (TP + TN + FP + FN+epsilon)
+    Prec = TP / (TP + FP+epsilon)
+    sen = TP / (TP + FN+epsilon)
+    F1_score = 2 * (Prec * sen) / (Prec + sen+epsilon)
+    CSI = TP / (TP + FN + FP+epsilon)
+    POD = TP / (TP + FN+epsilon)
+    FPR = FP / (FP + TN+epsilon)
+    FAR = FP / (TP + FP+epsilon)
 
     return (
         unique_values,
@@ -40,5 +40,4 @@ def evaluationmetrics(out_image1, out_image2):
         FPR,
         merged,
         FAR,
-        Dice,
     )