update model

@@ -107,7 +107,7 @@ class F3Classification(BaseModel):
         image = applications.mobilenet_v2.preprocess_input(image)
         return image, label
 
-    def load_dataset(self, dataset_dir, name, batch_size=128, augmentation_methods=[]):
+    def load_dataset(self, dataset_dir, name, batch_size=128, augmentation_methods=[], drop_remainder=True):
         image_and_label_list = self.get_image_label_list(dataset_dir)
         tensor_slice_dataset = tf.data.Dataset.from_tensor_slices(image_and_label_list, name=name)
         dataset = tensor_slice_dataset.shuffle(len(image_and_label_list[0]), reshuffle_each_iteration=True)
@@ -122,7 +122,7 @@ class F3Classification(BaseModel):
             self.preprocess_input, num_parallel_calls=tf.data.AUTOTUNE, deterministic=False)
         parallel_batch_dataset = dataset.batch(
             batch_size=batch_size,
-            drop_remainder=True,
+            drop_remainder=drop_remainder,
             num_parallel_calls=tf.data.AUTOTUNE,
             deterministic=False,
             name=name,
@@ -144,7 +144,8 @@ class F3Classification(BaseModel):
         )
         x = base_model.output
         x = layers.Dropout(0.5)(x)
-        x = layers.Dense(256, activation='sigmoid', name='dense')(x)
+        # x = layers.Dense(256, activation='sigmoid', name='dense')(x)
+        x = layers.Dense(256, activation='relu', name='dense')(x)
         x = layers.Dropout(0.5)(x)
         x = layers.Dense(self.class_count, activation='sigmoid', name='output')(x)
         self.model = models.Model(inputs=base_model.input, outputs=x)
@@ -243,7 +244,8 @@ class F3Classification(BaseModel):
             batch_size=batch_size,
             augmentation_methods=[
                 'rgb_2_bgr'
-            ]
+            ],
+            drop_remainder=False,
         )
 
         label_true_list = []