major bug fix

segmentation_models_trainer/dataset_loader/dataset.py

@@ -165,14 +165,20 @@ def process_csv_entry(entry):
                         separator=''
                     )
             )
-            label = decode_img(label, width, length, channels=1)
+            label = decode_img(
+                label,
+                width,
+                length,
+                channels=1,
+                interpolation_method=tf.image.ResizeMethod.NEAREST_NEIGHBOR
+            )
             # load the raw data from the file as a string
             img = tf.io.read_file(
-                entry['label_path'][0] if self.base_path == '' \
+                entry['image_path'][0] if self.base_path == '' \
                     else tf.strings.join(
                         [
                             self.base_path,
-                            entry['label_path'][0]
+                            entry['image_path'][0]
                         ],
                         separator=''
                     )
@@ -182,13 +188,13 @@ def process_csv_entry(entry):
             return img, label
 
 
-        def decode_img(img, width, length, channels=3):
+        def decode_img(img, width, length, channels=3, interpolation_method=tf.image.ResizeMethod.BILINEAR):
             # convert the compressed string to a 3D uint8 tensor
             img = tf.image.decode_png(img, channels=channels)
             # Use `convert_image_dtype` to convert to floats in the [0,1] range.
             img = tf.image.convert_image_dtype(img, IMAGE_DTYPE[self.img_dtype])
             # resize the image to the desired size.
-            return tf.image.resize(img, [width, length])
+            return tf.image.resize(img, [width, length], method=interpolation_method)
 
         def prepare_for_training(ds, batch_size):
             if self.cache: