dlr: tf graph changes

aimodel/src/deeplabv3_plus_test_rainfall.py

@@ -19,9 +19,9 @@ import tensorflow as tf
 from lib.dataset.dataset_mono import dataset_mono
 from lib.ai.components.LossCrossEntropyDice import LossCrossEntropyDice
 from lib.ai.components.MetricDice import metric_dice_coefficient as dice_coefficient
-from lib.ai.components.MetricSensitivity import sensitivity
+from lib.ai.components.MetricSensitivity import make_sensitivity as sensitivity
 from lib.ai.components.MetricSpecificity import specificity
-from lib.ai.components.MetricMeanIoU import one_hot_mean_iou as mean_iou
+from lib.ai.components.MetricMeanIoU import make_one_hot_mean_iou as mean_iou
 
 time_start = datetime.now()
 logger.info(f"Starting at {str(datetime.now().isoformat())}")
@@ -188,8 +188,8 @@ if PATH_CHECKPOINT is None:
 		metrics=[
 			"accuracy",
 			dice_coefficient,
-			mean_iou,
-			sensitivity, # How many true positives were accurately predicted
+			mean_iou(),
+			sensitivity(), # How many true positives were accurately predicted
 			specificity # How many true negatives were accurately predicted?
 			# TODO: Add IoU, F1, Precision, Recall,  here. 
 		],

aimodel/src/lib/ai/components/MetricMeanIoU.py

@@ -3,7 +3,9 @@ import math
 import tensorflow as tf
 
 
-def one_hot_mean_iou(y_true, y_pred, classes=2):
+def make_one_hot_mean_iou():
+	iou = tf.keras.metrics.MeanIoU(num_classes=classes)
+	def one_hot_mean_iou(y_true, y_pred, classes=2):
 		"""Compute the mean IoU for one-hot tensors.
 		Args:
 			y_true (tf.Tensor): The ground truth label.
@@ -20,7 +22,7 @@ def one_hot_mean_iou(y_true, y_pred, classes=2):
 		print("DEBUG:meaniou y_pred.shape AFTER", y_pred.shape)
 		print("DEBUG:meaniou y_true.shape AFTER", y_true.shape)
 		
-	
-	iou = tf.keras.metrics.MeanIoU(num_classes=classes)
+		iou.reset_state()
 		iou.update_state(y_true, y_pred)
 		return iou.result()
+	return one_hot_mean_iou

aimodel/src/lib/ai/components/MetricSensitivity.py

@@ -2,7 +2,10 @@ import math
 
 import tensorflow as tf
 
-def sensitivity(y_true, y_pred):
+
+def make_sensitivity():
+	recall = tf.keras.metrics.Recall()
+	def sensitivity(y_true, y_pred):
 		print("DEBUG:sensitivity y_pred.shape BEFORE", y_pred.shape)
 		print("DEBUG:sensitivity y_true.shape BEFORE", y_true.shape)
 		y_pred = tf.math.argmax(y_pred, axis=-1)
@@ -11,6 +14,8 @@ def sensitivity(y_true, y_pred):
 		print("DEBUG:sensitivity y_pred.shape AFTER", y_pred.shape)
 		print("DEBUG:sensitivity y_true.shape AFTER", y_true.shape)
 		
-	recall = tf.keras.metrics.Recall()
+		recall.reset_state()
 		recall.update_state(y_true, y_pred)
 		return recall.result()
+	
+	return _sensitivity